__group__ ticket summary component version milestone type owner status created _changetime _description _reporter Active Tickets 930 the legend keeps re-animating even if it is not changed other 2.0 bug Noughmad new 2011-08-23T12:32:57+02:00 2011-08-23T12:32:57+02:00 If I change some parameters in scatterplot widget, the legend is needlessly redrawn (it stays the same). marko Active Tickets 1172 sip4 crash on Ubuntu 10.10 when trying to use Visualization widgets library bug new 2012-04-04T04:02:28+02:00 2012-04-04T19:23:41+02:00 "I am trying out Orange for the first time on an Ubuntu 10.10 (Maveric) machine using Python 2.6. Whenever I connect a Visualization widget to a File widget, Orange immediately aborts with the error: {{{ python: /build/buildd/sip4-qt3-4.10.5/siplib/siplib.c:2600: sip_api_parse_result: Assertion `assign_helper != ((void *)0)' failed. Aborted }}} I get the same error whether I install Orange from debian packages or compile from source (Mercurial clone). I have spent several hours on this to no avail. I hope there is a fix because I would sure like to use visualization! Thanks. " pirobot Active Tickets 1189 On-line scripting reference should be cleaned up slightly documentation bug assigned 2012-05-28T08:18:48+02:00 2012-05-29T11:38:12+02:00 "I'm currently using Orange 2.5a4 but I've found a few problems with the documentation: * The online reference doesn't specify what version it applies to. I eventually found the PCA code I was looking for at `Orange.projection.pca.Pca` which is fine but confusing when the [http://orange.biolab.si/doc/reference/Orange.projection.linear docs] point me to `Orange.projection.linear.Pca`. I presume this is due to version mismatch, but without a version noted anywhere (at least that I could find) in the online reference, it was difficult to know to even look out for this. * The [http://orange.biolab.si/doc/reference/genindex/ reference index] is broken -- all links point to 404s * The PyPi page links to [http://orange-bioinformatics.readthedocs.org] which mostly unpopulated and therefore not very useful. It would be better to point to the documentation on the main site, or keep the reasdthedocs.org page updated" admackin Active Tickets 1190 Segfault when displaying data instances with discrete features library bug janez assigned 2012-05-29T04:09:43+02:00 2012-05-29T10:54:13+02:00 "I get a segmentation fault against 2.5rc1 when I try and display an instance which uses discrete features with no set of allowed values specified: {{{ >>> dom = data.Domain(feature.Discrete('a'), False) >>> inst = data.Instance(dom, [-1]) >>> print inst Segmentation fault }}} Specify a set of allowed values, and it works OK though: {{{ >>> from Orange import data, feature >>> dom2 = data.Domain(feature.Discrete('a', values=['y', 'n']), False) >>> inst2 = data.Instance(dom2, ['n']) >>> print inst2 ['n'] }}} " admackin Active Tickets 1193 Sesutje ob uporabi net file-a library bug miha assigned 2012-06-05T23:42:16+02:00 2012-07-26T10:36:16+02:00 "aplikacija se pa je sesula ko sem v kliknil na Net File, al pa ce v netfile-u ko ze imam nalozen net file hocem izbrati drugega (tud ce je tisti, ki pride skup z orangom) crashlog v prilozeni datoteki" maticko5 Active Tickets 1196 Data table segfault library bug janez new 2012-06-13T14:56:36+02:00 2012-06-13T14:56:36+02:00 "While the following code is not OK, it still should not result in a segmentation fault. {{{ import Orange data = Orange.data.Table('iris') dom = Orange.data.Domain(data.domain.features, False) data2 = Orange.data.Table(dom, data) cls2 = Orange.feature.Discrete(values=['0', '1'], name='class') cls2.get_value_from = lambda ins, rt: 0 dom.class_var = cls2 Orange.data.Table(dom, data2) }}} " lanz Active Tickets 1205 TypeError in Select Attributes on Orange Canvas canvas bug ales assigned 2012-06-22T20:38:09+02:00 2012-07-09T15:58:17+02:00 "When using the the Select Attributes module on Orange Canvas, I got the following error: {{{ There were problems importing the following widgets: OWLiftCurve: name 'AROCFromCDT' is not defined OWRScript: No module named rpy2.robjects OWModelMapQt: No module named orangeqt OWDiscretizeQt: No module named orangeqt OWDistributionsQt: No module named orangeqt OWLinProj3D: No module named orangeqt OWLinProjQt: No module named orangeqt OWMDSQt: No module named orangeqt OWNxExplorer3D: No module named OWNxExplorerQt OWParallelCoordinatesQt: No module named orangeqt OWPolyvizQt: No module named orangeqt OWRadvizQt: No module named orangeqt OWScatterPlot3D: No module named orangeqt OWScatterPlotQt: No module named orangeqt OWSphereviz3D: No module named orangeqt The following widgets could not be imported and will not be available: OWSphereviz3D, OWLiftCurve, OWModelMapQt, OWParallelCoordinatesQt, OWLinProj3D, OWScatterPlotQt, OWNxExplorer3D, OWRadvizQt, OWMDSQt, OWLinProjQt, OWDistributionsQt, OWScatterPlot3D, OWDiscretizeQt, OWPolyvizQt. The following prototype widgets could not be imported and will not be available: OWRScript. Unhandled exception of type TypeError occured at 10:39:30: Traceback: File: OWBaseWidget.py, line 678 in processSignals Code: handler(value) File: OWDataDomain.py, line 478 in set_data Code: self.used_attrs[:] = attributes File: OWItemModels.py, line 199 in __setslice__ Code: self.__delslice__(i, j) File: OWItemModels.py, line 189 in __delslice__ Code: self.beginRemoveRows(QModelIndex(), i, j - 1) TypeError: QAbstractListModel.beginRemoveRows(QModelIndex, int, int): argument 3 has unexpected type 'long' }}} The error occurs to every operation related to the ""Select Attributes"" module. Explored other modules on the Canvas, no same problem found so far. * Hardware: * Lenovo ThinkCenter * Intel i7-2600 * 16G mem * Software: * Windows 7 64bit * Python 2.7.3 64bit * Orange 2.5a4 was installed from ""Unofficial Windows Binaries for Python Extension Packages (http://www.lfd.uci.edu/~gohlke/pythonlibs/)"". I cannot use the official download because I need 64bit Python and other related Python modules. * Numpy: numpy-MKL-1.6.2.win-amd64-py2.7 * PyQt: PyQt-Py2.7-x64-gpl-4.8.6-1 * PyQwt: PyQwt-5.2.0-py2.7-x64 * PythonWin: pywin32-217.win-amd64-py2.7 I like Orange, which features and usability are very impressive! * It will be great if Orange can provide official 64bit Windows build (full packages with dependent libraries). * If the above is not possible due to license issue, it will be also good to provide standalone installation package on Windows, e.g. like the Orange Mac OSX dmg package, so it will install Python and other libraries in a separated location and not register Python in the system, which causing multiple Python in the system. * Where can I find a list of all Orange dependencies? " yangfd Active Tickets 1216 Import cycles addon/bioinformatics bug new 2012-07-17T14:09:20+02:00 2012-07-17T14:09:20+02:00 "We have import cycles in orange-bioinformatics and maybe somewhere else too. For example, one cannot import the following module: {{{ >>> import Orange.bio.obiPPI Traceback (most recent call last): File """", line 1, in File ""/home/lan/dev/orange/add-ons/Bioinformatics/_bioinformatics/obiPPI.py"", line 448, in from .obiGeneMania import wget, copyfileobj File ""/home/lan/dev/orange/add-ons/Bioinformatics/_bioinformatics/obiGeneMania.py"", line 489, in from . import obiPPI ImportError: cannot import name obiPPI }}} " lanz Active Tickets 1218 Fix for connecting to postgres server library bug new 2012-07-25T22:09:06+02:00 2012-07-25T22:09:06+02:00 "Connecting to a postgres server (9.1.4) fails on Fedora 17: {{{ >>> from Orange.data.sql import * /usr/lib64/python2.7/site-packages/Orange/network/__init__.py:59: UserWarning: Warning: some features are disabled. Install networkx to use the 'Orange.network' module. warnings.warn(""Warning: some features are disabled. Install networkx to use the 'Orange.network' module."") >>> r=SQLReader() >>> r.connect(""postgres://user:pass@127.0.0.1"") Traceback (most recent call last): File """", line 1, in File ""/usr/lib64/python2.7/site-packages/Orange/data/sql.py"", line 153, in connect (self.quirks, self.conn) = _connection(uri) File ""/usr/lib64/python2.7/site-packages/Orange/data/sql.py"", line 119, in _connection return (quirks, dbmod.connect(**dbArgDict)) File ""/usr/lib64/python2.7/site-packages/psycopg2/__init__.py"", line 179, in connect connection_factory=connection_factory, async=async) psycopg2.OperationalError: invalid connection option ""passwd"" >>> r.connect(""postgres://user@127.0.0.1"") Traceback (most recent call last): File """", line 1, in File ""/usr/lib64/python2.7/site-packages/Orange/data/sql.py"", line 153, in connect (self.quirks, self.conn) = _connection(uri) File ""/usr/lib64/python2.7/site-packages/Orange/data/sql.py"", line 119, in _connection return (quirks, dbmod.connect(**dbArgDict)) File ""/usr/lib64/python2.7/site-packages/psycopg2/__init__.py"", line 179, in connect connection_factory=connection_factory, async=async) psycopg2.OperationalError: invalid connection option ""db"" }}} I fixed this my modifying the postgres case in Orange/data/sql.py as follows: {{{ if schema == 'postgres': import psycopg2 as dbmod argTrans[""database""] = ""database"" #changed argTrans[""password""]=""password"" #added quirks = __PostgresQuirkFix(dbmod) quirks.parameter = ""%s"" }}} " bsloane Active Tickets 1220 ODBC/SQL Server connector library bug new 2012-08-01T04:23:43+02:00 2012-08-01T04:23:43+02:00 Extend the SQL interface to allow users to connect to Microsoft SQL Server databases. djstates Active Tickets 1224 .arff format transformation other bug new 2012-08-11T23:31:50+02:00 2012-08-11T23:31:50+02:00 "When saving dataset in .arff format, the relation name with more than one word should be within apostrophes. For example first line of the dataset: @relation Word Type Should be like: @relation 'Word Type'" szitnik Active Tickets 1225 Only one decimal after feature constructor widget canvas bug ales assigned 2012-08-14T12:58:27+02:00 2012-08-27T18:25:59+02:00 Attributes generated by the feature constructor widget are shown and output with only 1 decimal, which can be highly insufficient in many cases. The numbers are kept internally with higher precision, but there is no easy way in the canvas to override the formatting for display and file output. phusen Active Tickets 1227 String attributes export to Weka .arff library bug new 2012-08-16T20:14:59+02:00 2012-08-16T20:14:59+02:00 "Type of the attribute should be of type ""string"", not ""real"" at the beginning of the exported file." szitnik Active Tickets 1228 Empty string attribute values at Weka .arff export library bug new 2012-08-16T20:21:40+02:00 2012-08-16T20:21:40+02:00 "When empty string attribute value """" is exported, parentheses should be created ""''""" szitnik Active Tickets 1230 Table.save() method missing from the documentation documentation bug janez assigned 2012-09-04T14:22:11+02:00 2012-09-06T20:00:05+02:00 Table.save() method is missing from the documentation of Orange.data.table/ lefterav Active Tickets 1235 Orange Canvas will not run canvas bug new 2012-10-16T19:52:45+02:00 2012-10-19T10:54:15+02:00 "I am using Windows XP, Python 2.7. I have installed the latest version of Orange. When I double click on the Orange Canvas icon nothing happens. If I try to run from the python command line I get the following error: Traceback (most recent call last): File ""C:\Python27\Lib\site-packages\Orange\OrangeCanvas\orngCanvas.pyw"", line 15, in import orngTabs, orngDoc, orngDlgs, orngOutput, orngHelp, OWReport File ""C:\Python27\Lib\site-packages\Orange\OrangeCanvas\orngTabs.py"", line 12, in import orngHelp File ""C:\Python27\Lib\site-packages\Orange\OrangeCanvas\orngHelp.py"", line 3, in from PyQt4.QtWebKit import * ImportError: DLL load failed: The operating system cannot run %1. I have done many iterations of removing the PyQt program from the Control Panel and deleting the PyQt4 folder, and then re-running the Orange installer and it has not helped. I have also uninstalled the PyQt and then run a separate installer of PyQt4 from sourceforge.net. After doing this the Orange Canvas works, but I only get three visualization widgets. Can you help me to get the Orange Canvas working with all the visualization widgets?" efm Active Tickets 1241 Executing long-running python scripts hang the orange-canvas UI canvas bug new 2012-11-14T10:51:04+01:00 2013-01-07T14:59:25+01:00 "Steps to reproduce: Add a python script node with: {{{ while True: pass }}} Now try to move the node in the canvas. I'd suggest running every such codebyte in a separate thead or process. A different process would be nice because that way exit() won't kill orange-canvas. If this would cause too much overhead for data transfer then the thread option has another problem: E.g. try to use this python node: {{{ import thread def func(): print(123) thread.start_new_thread(func, ()) }}} But nothing is printed! See this error: {{{ Calling with ('123',) {} from the wrong thread. Queuing the call! Calling with ('\n',) {} from the wrong thread. Queuing the call! }}} So something went wrong in the queuing I guess. " ubershmekel Active Tickets 1243 Segmentation fault when training a randomforest learner library bug jurezb assigned 2012-11-16T02:04:30+01:00 2012-11-22T20:54:13+01:00 " I installed a few days ago using easy_install, so I guess it's the latest version of Orange. When learning RandomForestLearner, python raised an error ""Segmentation Fault"" and stopped. This crash happens on my window machine (Windows 7 x64) and on my linux machine (Sorry I don't know much about the linux machine) as well. Please find the attached an example script that crashes, and training/testing data." blisszen Active Tickets 1257 Documentation for building orangeqt with setup.py is missing canvas bug ales new 2012-12-12T11:26:29+01:00 2012-12-12T11:26:29+01:00 It should be included in INSTALL.txt in the source root ales Active Tickets 1284 NMF not working library bug MarinkaZitnik assigned 2013-02-15T12:43:07+01:00 2013-02-18T10:43:28+01:00 "the NMF library does not install. Obvisously the dependency to NIMFA is missing. Can you fix thisn please. " lareooreal Active Tickets 1285 Installer doeas not work - manual installation yields QT4 incompatibility library bug ales assigned 2013-02-15T13:04:44+01:00 2013-02-21T21:13:47+01:00 "The installationprocedure of Orange is ever since not working very well! Right now it requires manual installation of all components. Would you please mind to post a full list of requirements of _working_ software libraries of numpy,scypy etc.. Obviously the current QT4 version does not install and is not compatible. the plugins do not load all properly either. Please TEST your software before relasing- otherwise it is a lot of wasted time to try to setup a running installation. Greetings Lareo " lareooreal Active Tickets 1292 Can't Load Text Add On library bug new 2013-03-19T19:29:13+01:00 2013-03-19T19:29:13+01:00 "I keep trying to load the text add on and I get the following error just before it completes: error: unable to find vcvarsall.bat" gviviano Active Tickets 1293 SVMClassifier.get_binary_classifier raises a IndexError library bug ales new 2013-03-20T11:17:44+01:00 2013-03-20T11:17:44+01:00 "when the training set did not have any instances for a subset of classes (reported [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1812 here]) {{{ iris = Orange.data.Table(""iris"") iris_missing = Orange.data.Table(iris[50:]) # Remove all Iris-setosa instances svm_l = Orange.classification.svm.SVMLearner() svm_c = svm_l(iris_missing) # This breaks svm_c.get_binary_classifier(1, 2) }}} raises a {{{ --------------------------------------------------------------------------- IndexError Traceback (most recent call last) in () ----> 1 svm_c.get_binary_classifier(1, 2) /Users/aleserjavec/workspace/orange/Orange/classification/svm/__init__.pyc in get_binary_classifier(self, c1, c2) 584 (n_class - c2) 585 --> 586 coef = self.coef[classifier_i] 587 588 coef1 = [(mult * alpha, sv_i) for alpha, sv_i in coef \ IndexError: list index out of range }}} " ales Active Tickets 1296 Widgets and multitarget domains canvas bug ales new 2013-03-27T10:58:57+01:00 2013-03-27T10:58:57+01:00 "Many widgets strip the {{{class_vars}}} from the domain (select attributes, edit domain, ...). This makes using the Multitarget add-on difficult." ales Active Tickets 1297 Crash while loading C4.5 data other bug new 2013-04-14T00:17:04+02:00 2013-04-14T00:17:22+02:00 "Orange crashes while I attempt to load C4.5 data. It's deterministic. How to reproduce: 1. D&d ""File"" on a canvas 2. 2-click on ""File"" 3. Select *.data file (I tried to load files from C4.5 package) 4. Crash What I use: * C4.5 plugin build from currently available sources * orange-win-w-python-snapshot-hg-2013-04-13-py2.7 * Windows 8" ddr Active Tickets 1300 Double downloads in obiOMIM.py addon/bioinformatics bug vid new 2013-04-23T14:07:17+02:00 2013-04-23T14:07:17+02:00 "the morbidmap file is downloaded twice in obiOMIM.py (once in the constructor and once in the download_from_NCBI() method) is this really necessary?" vid Active Tickets 1302 Redirection in OMIM disease links addon/bioinformatics bug vid new 2013-04-25T13:16:17+02:00 2013-04-25T13:16:17+02:00 "The omimGeneSets() function in obiGeneSets.py uses links to a site, that get redirected to another site. The links should be replaced not to include the redirection. " vid Active Tickets 1303 network add-on not shown after installation other bug miha new 2013-04-25T18:44:18+02:00 2013-04-25T18:44:18+02:00 "If I install the network add-on through the canvas dialog, the new category is not shown and I need to restart Orange to see it, while Bioinformatics and NMF appear right after installation. It works the same in both Linux and Mac (bundle). Test with the old interface as the support for add-ons for the new interface is not completed." marko Active Tickets 1306 Fix global widget settings persistence canvas bug ales new 2013-05-22T19:18:54+02:00 2013-05-22T19:18:54+02:00 They should NOT be saved in files named by their title (possible name clashes, mismatch between the default 'title' parameter and widget meta description name loads from a filename never actually saved, ...) ales Active Tickets 870 Publish photos and posters from ISMB on Facebook web task gregor new 2011-07-29T13:22:42+02:00 2011-07-29T13:23:09+02:00 Publish photos, posters and short report from ISMB on Facebook (Biolab page). mitar Active Tickets 872 Follow data mining poll at KDnuggets other 2.0 task blaz assigned 2011-07-29T13:31:53+02:00 2012-05-29T11:02:42+02:00 "[http://www.kdnuggets.com/ KDnuggets] has a [http://www.kdnuggets.com/polls/index.html yearly Data Mining / Analytic Tools Used] poll. It starts somewhere in the May. We should follow it and make sure we rank food. " mitar Active Tickets 881 Check and maybe apply for Google Faculty Research Awards program other task blaz assigned 2011-08-03T18:32:28+02:00 2012-05-29T11:02:56+02:00 "Google provides [http://research.google.com/university/relations/research_awards.html university awards] for research in fields in which people working on Orange also do research. So we could check and maybe try to apply for some funding and research collaboration with Google. Deadline is April 15 2012." mitar Active Tickets 1113 AttributeError: ExampleGenerator.filter expects a list of conditions or orange.Filter other 2.0 task new 2012-02-28T14:05:22+01:00 2012-02-28T14:05:22+01:00 "Hi, i want to create a filter dinamically in runtime, as in example below, with the difference that the filter is now static. data = Orange.data.Table(""example.tab"") data2 = data.filter(sex=""female"") Can you help me? Thanx " peter Active Tickets 1254 Recent 'Documentation data sets' canvas task ales new 2012-11-26T18:23:07+01:00 2013-06-14T13:16:53+02:00 must be stored by name only not full path. This is necessary for tutorials style schemes to work, otherwise the filenames get filtered out. ales Active Tickets 1108 Fix data_files installation other 2.0 wish ales new 2012-02-23T16:17:38+01:00 2012-02-23T16:17:38+01:00 data files (docs, configuration files) should be installed using data_files and not package_files) ales Active Tickets 1112 MultitargetEntropy library 2.0 wish lanz new 2012-02-27T13:40:02+01:00 2012-02-27T13:40:02+01:00 "Add a feature scoring method for multi-target data with non-binary categorical classes to complement MultitargetVariance. Make sure MultiTreeLearner uses the appropriate score depending on data. Actually it would have to use a combination because of possible mixed (cont. & categ.) classes." lanz Active Tickets 1163 Python 3 support library wish new 2012-03-28T20:36:54+02:00 2012-03-28T20:36:54+02:00 There was [http://orange.biolab.si/forum/viewtopic.php?f=5&t=973 a forum post] suggesting python 3 support. Since numpy already has python 3 support, porting Orange should be less painful. varepsilon Active Tickets 1177 Orange Schema canvas wish marko assigned 2012-04-13T17:11:19+02:00 2012-06-29T14:40:21+02:00 "The current orange schema is far too sensible to dragging . Once all widgets are removed from the schema the schema does not come back to its original size. Also , it would nice to limit and autoresize the schema according to the placement of widgets . The required patch : go to => Orange/Orange/OrangeCanvas/orngView.py import time module and add to (mouseMoveEvent)line 279: time.sleep(0.01) This will reduce the sensibility of dragging. " romzee Active Tickets 1191 Widget for non-negative factorization addon/bioinformatics wish new 2012-05-29T21:52:22+02:00 2012-05-29T21:52:22+02:00 "Non-negative factorization (NMF) is a powerful statistical learning technic, packages already exist in Python. I would like to integrate this method into Orange (make an ""NMF widget""), under the supervision of Stan Young (National Institute of Statistical Sciences). Is this already planned by the Orange development team? If it is the case, it would be great to collaborate on this work. Please contact me for further collaboration if possible. Fajwel Fogel" Fajwel Active Tickets 1222 "display line number on C4.5 ""invalid length of example"" error" library wish new 2012-08-02T09:08:51+02:00 2012-08-02T09:08:51+02:00 "When loading a C4.5 file, I get the error ""invalid length of example"". It would be useful in debugging to display the line number on which this error occurred." kgeis Active Tickets 1223 Orange GUI for ETL jobs? other wish askeluv new 2012-08-08T15:00:37+02:00 2012-08-08T15:00:37+02:00 "I absolutely love the GUI for making schemes in Orange, so I was wondering if you would ever consider making the GUI ""detachable"" or if you'd be willing to team up with some people in order to make an Orange ETL tool..." askeluv Active Tickets 1286 Ctrl + Q shortcut new-canvas wish new 2013-02-20T12:42:21+01:00 2013-02-20T12:42:21+01:00 There should be a default shortcut for quit: Ctrl+Q (for linux and win). lanz Active Tickets 1298 Middle mouse button for scrolling canvas canvas wish new 2013-04-14T14:14:33+02:00 2013-04-14T14:14:33+02:00 "The middle mouse button is used in many programs for scrolling the canvas. E.g. Inkscape and GIMP. Would anyone mind if I implemented this?" ubershmekel Active Tickets 1313 Expand functionality of groupby widget. library wish echlebek new 2013-06-12T19:41:02+02:00 2013-06-14T12:41:29+02:00 "I'd like to add some functionality to this widget. 1. The ability to create multiple groupings per attribute. 2. A per-attribute boolean property which causes the aggregator to only consider unique values. 3. The ability to group into lists and sets. The first and second ideas are fairly easy to achieve. The second idea would naturally necessitate an ungrouping widget to deal with attributes that are collection objects. I'm also not sure how well the Orange data model would handle these types of objects." echlebek Active Tickets 1199 Orange wickets within java library 2.5 bug sjaakvw new 2012-06-15T12:16:32+02:00 2012-06-15T12:16:32+02:00 "I want to use a couple of orange wickets within Java. I have tried to use PyInterpreter to access een Orange python script (OWDataSampleA.py) but every time i got messages like ""cannot import name orange"" when is it trying to interpret 'from . import orange' C:\Python27\Lib\site-packages\Orange\__init__.py line 5. I think it has something to do with the pyd file which it can't read or find. The PYTHONPATH I have set with sys.path.append for example interpreter.exec(""sys.path.append('C:\\Python27\\Lib\\site-packages\\Orange\\orng')"") Question: can these Orange script be used outside python within java. And if so how (and if it's possible the solution to the problem I confronted you with) greetings sjaakvw" sjaakvw Active Tickets 1204 Orange.evaluation.testing documentation documentation 2.5 bug anze new 2012-06-20T17:46:15+02:00 2012-06-20T17:46:15+02:00 The functionality of Orange.evaluation.testing should be described as functions that users call, not through some internal construct (in this case, the Evaluation class). marko Active Tickets 1215 organize the Downloads page web 2.5 bug marko new 2012-07-09T18:04:36+02:00 2012-07-09T18:04:58+02:00 The downloads contains packages for linux that have not been updated since February. Check if all packaging methods for Macs are still usable. marko Active Tickets 1072 packages: svn -> hg other 2.0 2.5 task mitar accepted 2012-02-04T11:16:09+01:00 2012-05-29T11:31:38+02:00 Distribution - modify all the package building scripts to use the bitbucket repository. Delegate tasks if needed. marko Active Tickets 1074 Orange.feature.selection other 2.0 2.5 task miha reopened 2012-02-04T11:23:48+01:00 2012-05-03T12:51:36+02:00 Rewrite according to 2.5 conventions. marko Active Tickets 175 OWROC bug orng_modules 2.0 2.6 bug tomaz new 2008-01-15T10:38:09+01:00 2011-03-30T12:32:15+02:00 "OWROC crkne v orngStat: {{{ File: OWROC.py in line 857 Function name: testSetsSelectionChange Code: g.setShowIterations(list) File: OWROC.py in line 324 Function name: setShowIterations Code: self.calcAverageCurves() File: OWROC.py in line 420 Function name: calcAverageCurves Code: (averageCurves, verticalErrorBarValues) = orngStat.TCverticalAverageROC(ROCS, self.VTAsamples) File: orngStat.py in line 939 Function name: TCverticalAverageROC Code: TPsum.append( TP_FOR_FP(FPsample, ROCS[i], npts[i]) ) ##TPsum = TPsum + TP_FOR_FP(FPsample, ROCS[i], npts[i]) File: orngStat.py in line 922 Function name: TP_FOR_FP Code: return INTERPOLATE(ROC[i], ROC[i+1], FPsample) Exception type: Exception value: list index out of range }}} primer vrednosti, kjer crkne:[[BR]] i=1[[BR]] ROC=[(0.0, 0.0, 1.0000000000000001e+301), (0.0, 1.0, 0.80089092254638672)] ne vem kaj bi moral v tem primeru funkcija vrniti[[BR]] za repliciranje buga lahko uproabis widgetDebugging[[BR]] commandline ukaz: ""debugWidgets.py classify4.py""" gregor Active Tickets 176 OWLiftCurve bug canvas 2.0 2.6 bug tomaz new 2008-01-15T10:42:50+01:00 2011-03-30T12:17:04+02:00 "liftcurve vrze ta exception: {{{ File: OWLiftCurve.py in line 345 Function name: results Code: self.calcAllClassGraphs() File: OWLiftCurve.py in line 284 Function name: calcAllClassGraphs Code: g.setTestSetData(self.dresSplitByIterations, cl) File: OWLiftCurve.py in line 50 Function name: setTestSetData Code: P, N, curves = self.computeCurve(isplit, self.targetClass) File: OWLiftCurve.py in line 35 Function name: computeCurve Code: return orngStat.computeLiftCurve(res, classIndex) File: orngStat.py in line 1106 Function name: computeLiftCurve Code: curve.append( ((TP+FP)/(P + N), TP, (f, FP/N)) ) Exception type: Exception value: float division }}} za repliciranje buga lahko uproabis widgetDebugging[[BR]] commandline ukaz: ""debugWidgets.py classify4.py""" gregor Active Tickets 364 Delete Widget Settings ne zbrise settingsov za add-on-e canvas 2.0 2.6 bug gregor new 2009-03-17T18:20:04+01:00 2009-03-17T18:20:04+01:00 "Pri instaliranem modulu za bioinformatiko (pri meni se ta zavihek imenuje Genomics) odprem Heat Map, nastavim npr. Gamma na neko vrednost, izberem Options -> Delete Widget Settings, zaprem Orange (ne shranim nicesar), ga spet odprem in odprem isti widget. Vrednost za Gamma je se vedno taka, kot je bila, settings niso brisani. Kot kaze, ce prav razumem, settingse brises samo za core widgets, ne pa za widgete v add-onih. Ali lahko prosim add-ons vkljucis v brisanje (ravno se zafrkavam z enim widgetom, ki je v bioinformatiki in ki se mi zaradi buga pri dolocenih nastavitvah upira odpreti)." blaz Active Tickets 366 links to update information page from obiTaxonomy documentation documentation 2.0 2.6 bug blaz new 2009-03-18T12:16:13+01:00 2009-03-18T12:16:13+01:00 "This now says: ""These are also the organisms for which the information files, such as Gene Ontology annotation and KEGG pathways will be pre-loaded on Orange server and made available through Orange Bioinformatics database update."" Links are missing that will tell how this update is achieved (currently, there is no documentation on this). It also says: ""If there is an organism you wish to include in this list, please contact the authors."" This should have a link to ""Contact"" on Orange's web site. " blaz Active Tickets 557 Podpora za lokacijo ikonic widgetov drugje kot v site-packages canvas 2.0 2.6 bug miha assigned 2010-02-21T20:02:02+01:00 2012-05-29T11:28:48+02:00 "Podpora za lokacijo ikonic widgetov drugje kot v site-packages, saj ni po Linux standardu, da bi te bile med Python kodo: According to the Filesystem Hierarchy Standard, architecture-independent files need to be placed within /usr/share instead. Lahko bi se naredilo podobno kot smo naredili za dokumentacijo, da se ob izgradnji paketa v `orngConfiguration.py` zapiše recimo `iconsPath` spremenljivka. (Trenutno imamo tako `datasetsPath`.)" mitar Active Tickets 573 v orngStat ne dela sens/spec orng_modules 2.0 2.6 bug blaz new 2010-03-21T09:04:26+01:00 2010-03-21T09:04:26+01:00 " {{{ 642 return [sens(cm) for cm in confm] 643 else: --> 644 tot = confm.TP+confm.FN 645 if tot < 1e-6: 646 import warnings }}} " blaz Active Tickets 592 negative R2 orng_modules 2.0 2.6 bug blaz new 2010-05-17T09:30:31+02:00 2010-05-17T09:30:31+02:00 "TestLearners imaju u sebi R2 (R-squared). Po svemu R2 bi trebao biti pozitivan broj. Zašto ja dobivam R2 negativne vrijednosti (npr. -0,61)?" blaz Active Tickets 689 Orange.classification.lookup.LookupLearner bug library 2.0 2.6 bug janez new 2011-02-03T13:14:27+01:00 2011-02-03T13:14:27+01:00 "monk = Orange.data.Table(monks-1.tab) lc = Orange.classification.lookup.LookupLearner(monk) print lc(monk[10]) lc = Orange.classification.lookup.LookupLearner(monk) print lc(monk[10]) # prints 2 different answers ... " lanz Active Tickets 692 Verjetno nedelujoca metoda MeasureAttribute_MSE() orng_modules 2.0 2.6 bug somebody new 2011-02-28T22:32:45+01:00 2011-02-28T22:32:45+01:00 "V starejsi (pre-prenova) razlicici Orange mi je brez vecjih tezav delala metoda `orange.MeasureAttribute_MSE()` na tako diskretnih kot zveznih podatkih. Sedaj pa ne dela vec nikjer. Nisem se zelo poglobil v to kaj tocno potrebuje in kaksne so spremembe, a testiranje na `servo.tab`, `housing.tab` in razlicnih umetno generiranih domenah je vedno vrnilo NaN." gw.orange@… Active Tickets 718 PIPA widget and friendly authentication canvas 2.0 2.6 bug marko new 2011-04-01T18:25:32+02:00 2011-04-01T18:25:32+02:00 "In PIPA Widget the user is not sure if he's properly logged in or not. ""Authentication"" window only displays username and password, but does not say if the user is actually logged in. " blaz Active Tickets 726 ExampleTable_getslice crashes for negative indices library 2.0 2.6 bug janez new 2011-04-07T15:29:23+02:00 2011-04-07T15:29:23+02:00 "It returns a table containing invalid pointers. (The bug has been here since beginning, it's weird nobody noticed.) Also check all other _getslice methods." janez Active Tickets 859 Mushroom dataset je falsh. documentation 2.0 2.6 bug marko assigned 2011-07-08T16:16:52+02:00 2011-07-13T13:26:37+02:00 "V zbirki datasetov je ""mushroom.tab"" pokvecen: target class je nastavljen na zadnji atribut, namesto na prvi, ki manjka - imena atributov so temu primero zamaknjena, pri cemer je zadnje ime sproducirano iz predzadnjega...skratka, za popravit je. Zadeva je precej ocitna, saj daje CA okrog 50% namesto okrog 99%, tako da kake hujse skode od tega (upam ;) ) doslej ni bilo." MartinZ Active Tickets 946 When building both -O2 and -O3 flags are used other 2.0 2.6 bug ales new 2011-09-11T05:09:43+02:00 2011-09-11T05:09:43+02:00 When building (at least on Debian) both -O2 and -O3 flags are used. This should be made less confusing/cleaned/removed. mitar Active Tickets 950 Compiling of liborange_include.a must not use -fPIC other 2.0 2.6 bug ales reopened 2011-09-11T11:37:42+02:00 2011-09-20T11:17:24+02:00 By [http://www.debian.org/doc/debian-policy/ch-files.html#s-libraries Debian policy] static libraries must not use -fPIC when compiling (but dynamic must). We are compiling `liborange_include.a` statically. mitar Active Tickets 994 Fink install on Mac OS X Lion other 2.0 2.6 bug anze assigned 2011-10-28T12:10:54+02:00 2011-11-05T00:15:25+01:00 Install script [source:trunk/install-scripts/mac/fink-selfupdate-orange.sh] does not work on Mac OS X Lion. anze Active Tickets 1001 debian packege: networkx to recommends other 2.0 2.6 bug marko assigned 2011-11-02T15:26:16+01:00 2011-11-05T01:09:30+01:00 Please move networkx from dependencies to recommends. marko Active Tickets 1030 Errors in documentation documentation 2.0 2.6 bug marko new 2011-12-13T21:01:27+01:00 2011-12-13T21:11:10+01:00 "There are some errors in documentation: * undefined label `file-formats` is used * file `code/variable-reuse.py` is missing * in `associate/__init__.py` there are hard-coded links to `Domain.htm`, `Value.htm` and `fileformats.htm` * `feature/discretization.py` has a hard-coded link to `o_categorization.htm` * `feature/imputation.py` has a hard-coded link to `callbacks.htm`" mitar Active Tickets 1118 Lasso Regression is slow on auto-mpg dataset. library 2.0 2.6 bug lan new 2012-03-05T13:09:26+01:00 2012-05-04T18:27:20+02:00 "Learning of Lasso Regression learner on auto-mpg dataset is very slow. It also generates division by zero warnings. Please check and fix if possible. code to reproduce (run python as ""python -Wall"" to enable warnings): {{{ import Orange auto = Orange.data.Table(""auto-mpg"") lr = Orange.regression.lasso.LassoRegressionLearner(n_boot=2, n_perm=2) lr(auto) }}} " anze Active Tickets 1155 Statistics in linear regression library 2.6 bug lan new 2012-03-26T14:57:28+02:00 2012-03-26T14:57:28+02:00 "The following still has to be checked/corrected: - for unweighted data, should the std (and the correction in r2adj) be computed using n instead of (n-1) when intercept=False - for weighted data, what is the correct computation of F and r2" lanz Active Tickets 1175 NaiveLearner on continuous features library 2.6 bug anze new 2012-04-11T14:28:22+02:00 2012-04-11T14:28:22+02:00 "NaiveLearner works on continuous features, but not when they are all 0 or 1. Probably needs some adjustment in loess window settings. To reproduce: {{{ import Orange data = Orange.data.Table('voting') continuizer = Orange.data.continuization.DomainContinuizer() continuizer.class_treatment = continuizer.Ignore domain0 = continuizer(data) data0 = data.translate(domain0) c = Orange.classification.bayes.NaiveLearner(data0) print c.conditional_distributions }}} Compare for example to using it on iris, where it works fine." lanz Active Tickets 1197 Inconsistent SVMClassifier library 2.6 bug ales new 2012-06-14T13:43:06+02:00 2012-06-14T13:43:06+02:00 "{{{ classifier(inst,classifier.GetBoth)[1] classifier(inst) }}} can return different values in case of ties [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1611 see] This should be fixed by overriding the {{{TClassifier::predictionAndDistribution}}} method." ales Active Tickets 1203 DomainContinuizer and multitarget data library 2.6 bug ales new 2012-06-18T12:05:02+02:00 2012-06-18T12:05:02+02:00 "When DomainContinuizer is given a data set with multiple classes it returns a domain with continuized features, but without classes (class_vars is not copied from the original domain). {{{ data = Orange.data.Table('emotions') dom = Orange.data.continuization.DomainContinuizer()(data) print data.domain.class_vars, dom.class_vars }}} " lanz Active Tickets 1258 orangeqt sip build flags library 2.6 bug ales new 2012-12-13T15:37:56+01:00 2012-12-13T15:39:08+01:00 source:orange/source/orangeqt/setup.py falsely assumes PyQtw was build against the current Qt4 version [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1731 see] ales Active Tickets 1259 EarthLearner fails TestLearners precheck canvas 2.6 bug ales new 2012-12-19T11:21:30+01:00 2012-12-19T11:21:30+01:00 "If the domain size is small (less than 4). In line 326 {{{ n = len(self.data.domain.attributes)*2 indices = orange.MakeRandomIndices2(p0=min(n, len(self.data)), stratified=orange.MakeRandomIndices2.StratifiedIfPossible) new = self.data.selectref(indices(self.data), 0) }}} n can be too small, EarthLearner requires at least 8 instances. " ales Active Tickets 1264 Widgets taking an SymMatrix should not fail when it does not have a '.items' attribute canvas 2.6 bug ales new 2013-01-09T11:09:28+01:00 2013-03-19T12:41:57+01:00 The items attribute is not a builtin attribute of the SymMatrix. ales Active Tickets 865 Check how setup.py works other 2.0 2.6 check anze assigned 2011-07-23T09:37:01+02:00 2011-11-24T01:38:48+01:00 Check how setup.py works on different platforms. Can we use it everywhere to install Orange and generate packages for Fink/Debian and bundle? Currently we are using there custom scripts. mitar Active Tickets 1187 Naive Bayes in Orange or Weka library 2.6 check marko accepted 2012-05-14T14:55:00+02:00 2012-06-29T14:43:41+02:00 " The results obtained by Naive Bayes from Orange differ from the results obtained by Weka. For example on ""ZOO"" data set the CAs are:[[BR]] Orange: 92.12[[BR]] Weka: 93.06 Both uses Laplace for prior probability estimation. How I can repeat the experiments from Weka in Orange? " cankarm Active Tickets 260 redesignirati glavo v dokumentaciji widgeta documentation 1.0 2.6 task somebody new 2008-06-20T00:50:15+02:00 2011-03-10T00:44:14+01:00 Ni prav posebej lepa ;) Predstavljam si nek okvir. V njem je levo zgoraj je ikona widgeta, desno od nje ime in kratek opis. Desno od tega pa je seznam signalov. janez Active Tickets 470 preveri GO enrichment pri izbrani referencni mnozici canvas 2.0 2.6 task blaz new 2009-07-10T14:25:33+02:00 2009-07-10T14:25:33+02:00 primerjaj z davidom blaz Active Tickets 754 reference documentation - Attribute types defined in Python documentation 2.0 2.6 task assigned 2011-05-03T11:32:14+02:00 2012-02-14T11:26:25+01:00 marko Active Tickets 780 orngLinVis other 2.0 2.6 task new 2011-05-03T11:51:59+02:00 2012-02-07T21:59:27+01:00 marko Active Tickets 782 orngVisFuncts, orngVizRank other 2.0 2.6 task new 2011-05-03T11:52:34+02:00 2012-02-07T21:59:00+01:00 marko Active Tickets 842 Start using orangerc.cfg instead of orngConfiguration.py other 2.0 2.6 task miha assigned 2011-07-01T01:12:18+02:00 2012-05-29T11:29:51+02:00 Start using `orangerc.cfg` instead of `orngConfiguration.py` in building packages (Fink and Debian) to configure where example datasets (and icons) are. mitar Active Tickets 925 Build orangeqt in VM library 2.0 2.6 task miha assigned 2011-08-22T13:16:04+02:00 2012-05-29T11:25:56+02:00 Check if orangeqt compiles in VMs and if it is included in Mac OS X package. miha Active Tickets 962 Fedora package: prepare spec file other 2.0 2.6 task tadej new 2011-09-15T11:23:23+02:00 2011-09-21T15:26:01+02:00 Prepare spec file for building Fedora Orange packages. You can check [source:trunk/install-scripts/debian/control-files/rules Debian] and [source:trunk/install-scripts/mac/fink/orange-gui-svn-py.info Fink/Mac] packaging scripts. mitar Active Tickets 1047 orangeqt - make installation possible with setup.py other 2.0 2.6 task Noughmad assigned 2011-12-23T11:27:55+01:00 2012-04-13T20:17:05+02:00 "We are trying to make orange installable with setup.py, but as for now it does not support orangeqt. Ales hinted, that we will probably need to make a new setup.py for orangeqt." marko Active Tickets 1164 Documentation included in the install packages documentation 2.6 task ales new 2012-03-29T17:36:23+02:00 2012-04-18T18:00:58+02:00 Compiled and up to date documentation must be included in the install packages (Windows and Mac OS). ales Active Tickets 1200 GUI interface for installation of add-ons canvas 2.6 task ales new 2012-06-15T16:47:03+02:00 2012-06-15T17:34:46+02:00 "Now that we have add-ons on PyPi, it would be user friendly that people could install them through user interface in canvas. PyPi has nice API to check for our add-ons. Our add-ons have keyword `orange add-on`. [https://github.com/mitar/cmsplugin-pypipackages/blob/master/cmsplugin_pypipackages/management/commands/pypipackages.py Here is available code] I am using for display on the web page. For installation to be possible, egg should exist for the given platform (if add-on has binary extensions) or platform independent egg. So interface should list all add-ons and allow installing only those which are possible. Egg should be installed by simply adding it into the `add_ons_dir_user` directory. I am not sure if it is possible to dynamically reload add-ons (without Orange restart), but probably yes. Just run `Orange.utils.environ.load_eggs`." mitar Active Tickets 1201 Building of binary add-ons other 2.6 task miha new 2012-06-15T16:48:43+02:00 2012-06-15T17:34:53+02:00 "Some add-ons have binary extensions. They should be build into eggs for those platforms and published on PyPi. This should be done so that it can be done easily once we publish new version of our add-ons. " mitar Active Tickets 1210 get rid of server files server part - use something standard library 2.6 task marko new 2012-06-29T16:23:17+02:00 2012-06-29T16:23:17+02:00 marko Active Tickets 295 Survey Plot - Selection of examples canvas 2.0 2.6 wish gregor new 2008-08-11T15:59:31+02:00 2008-08-11T15:59:31+02:00 Survey Plot bi bilo morda lepo razsiriti tako, da bi lahko v njem izbral primere (drag za izbor podrocja, CTRL-CLICK & Drag za izbor disjunktnih podrocij). blaz Active Tickets 456 orngText reviwal orng_modules 2.0 2.6 wish blaz new 2009-05-29T10:26:35+02:00 2009-05-29T10:27:55+02:00 "Get rid of orngText.PreprocessorConstructor_tfidf()(data), write constructors to handle the data in initialization directly. Write orngText.tfidf (calls constructur and then a method, shortcut to above). Write more examples. Include zipped distance (move the implementation from the example python script in text directory). Change names from lemmatizeExampleTable to lemmatizeData, and similar." blaz Active Tickets 1013 Implement predict_all on base learner other 2.0 2.6 wish new 2011-11-29T13:16:04+01:00 2011-11-29T13:16:04+01:00 "Base Learner should implement method predict_all(Table data). Base implementation of this method would iterate through all instances in table and return list of predictions. Learners that extend classification.Learner could overload this method to provide more efficient implementation (i.e. when results are computed with multiplication of numpy matrices, one could multiply ""model"" matrix with example matrix and return results)." anze Active Tickets 1091 references in documentation addon/bioinformatics 2.0 2.6 wish blaz new 2012-02-08T10:19:24+01:00 2012-02-08T10:19:24+01:00 "Unify the reference style in the documentation. Improve the style file. For example, see http://orange.biolab.si/doc/reference/Orange.associate/." blaz Active Tickets 1092 print model addon/bioinformatics 2.0 2.6 wish new 2012-02-08T15:48:19+01:00 2012-02-08T15:48:19+01:00 "Check that all models inferred from the data canbe ""printable"", like: {{{ print model }}} Some already are, like linear regression." blaz Active Tickets 1125 Distributions (widget) canvas 2.0 2.6 wish new 2012-03-09T10:35:27+01:00 2012-03-09T10:35:27+01:00 "In widget called distributions the presentation for real-valued features should be replaced with a standard box plot. Like: [[Image(https://img.skitch.com/20120309-g74j3db79dqfa2hrqy3jagk73y.jpg)]]" blaz Active Tickets 1179 Widget highlighting canvas 2.6 wish blaz assigned 2012-04-19T09:26:33+02:00 2012-06-29T14:41:48+02:00 "Highlighting the compatible widgets to a particular widget when the mouse hovers over that widget in the schema, would be a fascinating and quite helpful experience in orange. Hey , here's a suitable design to the patch : add 2 more line to widget popup menu : ( orangeCanvbas.pyc:line 497) self.widgetPopup.addAction(""Common signal inputs"",self.input_types) self.widgetPopup.addAction(""Common signaloutputs"",self.output_types) #now define the two methods :(line 523) def input_types(self): """""" the code follows here """""" return def output_types(self): """""" the code follows here """""" return """""" On right clicking the widget , the options to find common input/output widgets comes in the pop up menu"""""" """""" The result of clicking this , will either highlight the widgets or popup a menu containing these widgets """""" " romzee Active Tickets 1229 TreeClassifier to_networkx_tree() function library 2.6 wish marko assigned 2012-08-27T23:22:26+02:00 2012-09-06T20:03:25+02:00 "The TreeClassifier has a to_string() function that prints the decision tree in text format. Even though this function can be customized via formatting arguments, the output is still unstructured and the tree cannot be easily traversed and inspected. Below, I propose to add a new function called to_netwrokx_tree() that converts the internal Orange tree structure to a Networkx directed graph, where internal nodes are features and leaves are decisions. Internal node labels are feature names and edges' labels are feature values on which the tree splits. The leaves' labels are the decisions along with their probabilities. {{{ def to_networkx_tree(pnode, tree=None, level=0, local_index=0): '''Returns a networkx tree of the orange classification tree''' # create a new directed graph if this is the first call if not tree: tree = networkx.DiGraph() # reached null nodes if not pnode: return tree # if the node has children (not leaves), traverse if pnode.branch_selector: pnode_desc = pnode.branch_selector.class_var.name # create a root node only on first call if level == 0: pnode_id = str(pnode_desc) + '_%i' % len(tree) tree.add_node(pnode_id, shape='none', label=str(pnode_desc)) else: pnode_id = str(pnode_desc) + '_%i' % (len(tree) - 1) # iterate through children and update branch for i in range(len(pnode.branches)): cnode = pnode.branches[i] # if the child is not a null node, update the tree if cnode: # if the child is a parent, update a branch if cnode.branch_selector: # add child node cnode_desc = cnode.branch_selector.class_var.name cnode_id = str(cnode_desc) + '_%i' % len(tree) tree.add_node(cnode_id, shape='none', label=str(cnode_desc)) # add edge between parent and child edge_desc = str(pnode.branch_descriptions[i]) tree.add_edge(pnode_id, cnode_id, label=edge_desc) # recursively update this branch of the tree tree = get_tree(cnode, tree, level+1, i) # if the child is a leaf, then add class info else: leaf_dist = cnode.node_classifier.default_distribution leaf_prob = leaf_dist[int(cnode.node_classifier.default_val)] leaf_class = cnode.node_classifier.default_value leaf_id = str(leaf_class) + '_%i' % len(tree) leaf_label = str(leaf_class) + ' (%.2f%%)' % (leaf_prob*100) tree.add_node(leaf_id, shape='box', label=leaf_label) # add edge between parent and child edge_desc = str(pnode.branch_descriptions[i]) tree.add_edge(pnode_id, leaf_id, label=edge_desc) # null node reached else: pass # once done with children for this parent, return tree return tree # if it's a leaf and it's a root, update the node with class info else: if level == 0: leaf_dist = pnode.node_classifier.default_distribution leaf_prob = leaf_dist[int(pnode.node_classifier.default_val)] leaf_class = pnode.node_classifier.default_value leaf_id = str(leaf_class) + '_%i' % len(tree) leaf_label = str(leaf_class) + ' (%.2f%%)' % (leaf_prob*100) tree.add_node(leaf_id, shape='box', label=leaf_label) return tree }}} Here's how the code could be used and tested: {{{ import orange, Orange import networkx import networkx.algorithms as nx_algorithms # use voting dataset to train a decision tree data = orange.ExampleTable('voting') tree_classifier = orange.TreeLearner(data) # convert the orange tree to networkx tree networkx_tree = to_networkx_tree(tree_classifier.tree) # assert that both trees have the same order, have one connected components, and is a DAG assert len(networkx_tree) == tree_classifier.tree.tree_size() assert nx_algo.number_connected_components(networkx_tree.to_undirected()) == 1 assert nx_algo.is_directed_acyclic_graph(networkx_tree) == True # convert the tree to AGraph and save it as a Graphviz DOT file agraph = networkx.to_agraph(networkx_tree) agraph.write('networkx_tree.dot') }}} " yazan.boshmaf Active Tickets 1291 Connect to a mysql using a unix_socket other 2.6 wish new 2013-03-12T16:13:08+01:00 2013-03-12T16:13:08+01:00 "Hello, I have just registered because I just downloaded orange and I would like to connect to a local mysql using a unix socket. But I couldn't. And I just take a look into your really nice code, and added two lines in Orange/data/sql.py to be able to connect to a mysql using a unix_socket. And I would like to share this lines with you: def _connection(uri): . . . if path: dbArgDict[argTrans['database']] = path[1:] ''' if schema == 'mysql' and args and args.has_key('unix_socket'): dbArgDict['unix_socket'] = args.get('unix_socket')''' return (quirks, dbmod.connect(**dbArgDict)) Those lines make posible to use a connection string like the next one and be able toconnect to the mysql database. 'mysql://root:@localhost/users?unix_socket=/usr/local/zend/mysql/tmp/mysql.sock' I don't know If you will consider to add this lines onto your code, but I will keep them on my local copy because help me to use orange. Thanks a lot for your work Jordi" jarnavat Active Tickets 1165 What does earth do if it encounters unknown class values? library 3.0 check new 2012-03-29T22:07:32+02:00 2012-04-03T01:19:19+02:00 Nothing, currently. It assumes that there are no missing values. janez Active Tickets 768 reference documentation - Subtyping Orange classes in Python documentation 2.0 3.0 task new 2011-05-03T11:42:00+02:00 2012-02-14T11:30:35+01:00 marko Active Tickets 863 Conversion constructors for Example library 2.0 3.0 task janez new 2011-07-20T22:16:55+02:00 2011-07-20T22:16:55+02:00 "Constructors orange.Example(domain, example) and orange.Example(domain, list-of-examples) need to be implemented. For now, they are omitted since conversion routines are not working yet -- we don't have classifiers and we can't have classifiers before we have examples." janez Active Tickets 873 Reimplement PythonVariables other 2.0 3.0 task janez new 2011-08-01T12:47:11+02:00 2011-08-01T21:30:32+02:00 "This ticket will list the places at which code related to PythonVariables needs to be reenabled: TDomain::domainCompatible TExampleTableReader::typeIdentifiers" janez Active Tickets 876 Optimized file reading library 2.0 3.0 task janez new 2011-08-01T16:24:05+02:00 2011-08-01T16:24:16+02:00 When reading files, values should be read into a memory first; this will speed up detection of attribute types and probably reduce the table size since exact line count will be available before allocating it. janez Active Tickets 936 Add unit tests for continuous domains for kNN and Majority other 2.0 3.0 task janez new 2011-09-02T15:27:20+02:00 2011-09-02T15:27:20+02:00 janez Active Tickets 958 Replace allVariables with a multimap library 2.0 3.0 task janez assigned 2011-09-12T23:52:32+02:00 2012-03-24T22:20:51+01:00 "For variables to match, they must have the same varType and name. Since all variables are, in principle, constructed through TVariable::make, lookups take O(n^2) when, say, loading a data file. This may be quite significant with 10000 variables. We can replace allVariables, which is currently a list, with a multimap indexed by varType and name. Alternatively (and perhaps better), we can have separate multimaps for each varType; multimaps key would then be a name." janez Active Tickets 960 Setting errors versus throwing exceptions other 2.0 3.0 task janez new 2011-09-15T00:07:12+02:00 2011-09-15T00:07:12+02:00 "As a general principle, functions that return PyObject * will set the error (e.g. PyErr_Format) and functions that return POrange or TOrange * will raise exceptions. Check that the code conforms to this (and fix it) ... or change the principle. ;) Rationale: say that some code calls ex = TExample::fromDomainAndPyObject(domain, item); Should it check that ex is not NULL or should it catch exceptions? Orange will use exceptions internally but functions closer to Python will use Python API for setting error messages. Although P- and Py-classes are switched back and forth, the point at which the former convert to the latter is suitable for also defining type of exceptions. Functions that deal with PyObjects still do not have to catch exceptions, although we need to ensure that any exceptions are eventually caught and ""Pythonized""." janez Active Tickets 970 Add unit tests for entropy discretization other 2.0 3.0 task janez new 2011-09-27T18:19:59+02:00 2011-09-27T18:19:59+02:00 janez Active Tickets 974 Faster threshold search other 2.0 3.0 task janez new 2011-09-29T20:04:47+02:00 2011-09-29T20:19:35+02:00 "Threshold search could be sped up by implementing special cases of procedures related to the search for binary attributes that would use a pair of double instead of TDistribution. An alternative is to just avoid TDistribution and use a double * instead. TScoreFeature would have to implement a corresponding method. Third alternative is to even keep TDistribution but only one (eg. the left one), implement TScoreFeature's operator that would accept the left distribution and the class distribution. This would simplify the threshold search loop (no maintining of the count for the right side) and possibly allow a more efficient computation of scores as well." janez Active Tickets 1068 Assigning vectors to tables library 2.0 3.0 task janez new 2012-01-30T14:42:36+01:00 2012-01-30T14:42:36+01:00 Allow data[5:10, 7] = [1, 2, 3, 2, 1] janez Active Tickets 1128 Unit tests for imputation library 2.0 3.0 task janez new 2012-03-10T09:58:01+01:00 2012-04-03T21:11:51+02:00 janez Active Tickets 1129 Unit tests for logistic regression fiter library 2.0 3.0 task janez new 2012-03-10T09:58:34+01:00 2012-03-10T09:58:34+01:00 janez Active Tickets 1138 """Smart new"" for Learner" library 2.0 3.0 task janez new 2012-03-13T14:09:14+01:00 2012-03-13T14:09:14+01:00 "Learner shouldn't be abstract. operator::() should call the method __call__ in Python __new__ should be smart: - construct learner - if there is no data: -- call __init__ (if needed) with all keyword arguments - else -- call learner's __init__ with keyword arguments with matching names OR with all keyword arguments if __init__ accepts **kw -- call __call__ -- set attributes corresponding to keyword arguments (none, if the learner's init accepts **kw)" janez Active Tickets 1142 Clsutering in Orange 3.0 needs more unit tests library 2.0 3.0 task new 2012-03-19T13:43:42+01:00 2012-03-19T13:43:42+01:00 janez Active Tickets 1151 Reimplement RemoveRedundantOneValue and RemoveUnused values in Python library 3.0 task new 2012-03-23T14:26:08+01:00 2012-03-23T14:26:08+01:00 janez Active Tickets 1162 ExamplesDistance_Relief and ignoreClass library 3.0 task new 2012-03-27T23:22:54+02:00 2012-03-27T23:22:54+02:00 "I encountered this when merging changes from 2.5. ReliefF (distance, not score!) seems to have been ignoring ignoreClass (it issued a warning regarding this); it computed only distances disregarding the class values. Then we added the class ... but not all the way: it computed the class statistics, but did not use it when computing distances. We also did not remove the warning about not ignoring class. Find out how should we treat the class (e.g. Kononenko's papers or books) and fix the code accordingly." janez Active Tickets 1169 Check unpickling of Variables library 3.0 task new 2012-03-30T21:34:33+02:00 2012-03-30T21:34:33+02:00 When merging from 2.5, changeset 8934 was skipped since pickling in Orange 3.0 stores all attributes in the state, so it should not have the problem addressed by this fix. Check it, that is, add a unit test that constructs a variables whose get_value_from refers to this same variable, pickle it, delete the variable (so unpickle does not reuse it) and unpickle it. janez Active Tickets 1174 Add more unit tests for ExampleTable.save library 3.0 task new 2012-04-06T23:45:23+02:00 2012-04-06T23:45:23+02:00 "Much of the code has been rewritten. Tests need to include - meta attributes - txt files that need to indicate that the attribute is discrete since it could be mistaken for continuous - basket format - weird examples, like txt or tab with only meta attributes, tab with only a basket column etc" janez Active Tickets 969 Consider making Contingency more efficient other 2.0 3.0 wish janez new 2011-09-27T17:04:11+02:00 2011-09-27T17:04:28+02:00 "In particular, in discrete-vs-discrete case it can be stored as double *, and in continuous-vs-discrete as map (instead of map). In either case, we will have problems with operator[] that returns a PDistribution. A possible solution is to allow TDiscDistribution::distribution to point to shared data. In that case, TDiscDistribution::distribution_ would be put in union with a bool telling whether the data is owned or borrowed. The question is how much would we gain by this optimization, in particular in discrete-vs-discrete case. Discrete contingency tables are typically small." janez Active Tickets 1150 Multi-dimensional class variables library 3.0 wish new 2012-03-22T11:49:57+01:00 2012-03-22T12:21:47+01:00 "Orange 2.5 domain can contain a single class variable (class_var) and/or multiple class variables (class_vars). It is possible (and common) to construct a domain that has class_var set and empty class_vars or vice versa. In Orange 3.0, we should merge the concepts of single-class and multi-class domains, since they both represent the dependent variables. * If a domain contains one or more class variables, class_vars should contain all of them. class_var should point to the first variable in class_vars. * data.Instance get_class should return the value of the first dependent variable or raise exception if domain is classless. * indexing instances with index -1 should raise a deprecation warning. If domain has a single class, it should return its value. If domain has more than one class, it should raise an error. * to_numpy should default to ""am/w"". Most of the methods that use it should use as_numpy anyway. " anze Active Tickets 31 Scatterplot: trendline Future bug gregor new 2007-05-02T11:32:37+02:00 2007-05-02T11:32:37+02:00 V Scatterplot bi bilo zanimivo dodati trendline. Ce uporabljas barve, dobis trend za vsako barvo, sicer pa enega za vse. Lahko bi dodali tudi konfidencne intervale. To rec lahko delas z linearno regresijo ali pa tudi z Loessom. janez Active Tickets 32 Data sample: Oznaka folda kot nov atribut Future bug new 2007-05-07T19:36:43+02:00 2007-05-07T19:36:43+02:00 Data sample bi lahko imel opcijo, da da ven vse podatke, vendar pripne fold kot nov atribut. janez Active Tickets 138 coloring of points in Correspondence Analysis orng_modules 1.0 Future bug somebody new 2007-11-29T16:46:47+01:00 2008-05-19T01:32:51+02:00 "Remove the use of ""category"" attribute, instead implement the pull-down menu with which the user decides on which coloring attribute to use" anonymous Active Tickets 139 CA shuld not use category attribute orng_modules 1.0 Future bug sasa new 2007-11-29T17:04:50+01:00 2008-05-19T01:32:02+02:00 instead make it work like scatterplot (user can decide which attribute to use to color the points). anonymous Active Tickets 201 CA, AUC ipd. pravilno upoštevajo uteži testnih primerov? orng_modules 1.0 Future bug blaz new 2008-03-25T22:54:41+01:00 2011-07-01T09:18:47+02:00 janez Active Tickets 289 Scatterplot - VizRank - Outliers ne vzame označenega primera canvas 1.0 Future bug gregor new 2008-08-07T10:45:19+02:00 2008-08-07T10:45:19+02:00 "Na scatterplotu oznacim en sam primer. V outlier identification pritisnem ""Get from Projection"", izpise se mi warning ""Exactly one example must be selected""." blaz Active Tickets 292 Mosaic Display - subboxes canvas 1.0 Future bug gregor new 2008-08-08T14:02:07+02:00 2008-08-08T14:02:07+02:00 "Namesto ""Use subboxes on the left to show..."" naj bo Display sub-box (check button) Expected class distribution / Apriory class distribution (pull down) obe kontroli naj gresta na vrh boxa (pred ostali dve kontroli). ""Show class distribution of subset data"" skrajsaj v ""Show subset data distribution"", in ga disable-aj ce ni data subset-a ni. (ce se da, spremeni in v verziji 1.0 in 2.0)" blaz Active Tickets 300 data set info files documentation 2.0 Future bug blaz new 2008-08-24T10:34:20+02:00 2008-08-24T10:34:20+02:00 "set all the data set info files so that they contain a link to the data set (this could require them to be set in XML). extend info files so that they contain a brief description of the data set as shown on the web site, and perhaps a few tags" blaz Active Tickets 305 GO: ob updatanju annotacij je treba posodobiti ontologijo orng_modules 2.0 Future bug ales new 2008-09-25T14:11:51+02:00 2008-09-25T14:11:51+02:00 "Ob downloadanju anotacij se mora posodobiti tudi Ontologija. Če vzamem novo anotacijo, ne updejtam pa ontologije, potem ne najde termov v ontologiji. " marko Active Tickets 320 F1 vcasih ne prime canvas 2.0 Future bug gregor new 2008-11-23T00:36:02+01:00 2008-11-23T00:36:02+01:00 "F1 v widgetih vcasih prime vcasih ne. Podobno ne deluje avtomatska sinhronizacija helpa (tako, da bi se stran sama zamenjala na widget, ki je trenutno na vrhu. Videti je, da se ne poklica focusInEvent za widget temvec, ugibam, za kontrolo. Najbrz je tudi prvi problem povezan s tem - morda F1 prileti v trenutno aktivno kontrolo, ne v widget. Gregor, bi mogoce lahko ti pogledal, v cem je hec - tele zadeve s signali so jasne kvecjemu tebi?" janez Active Tickets 322 attribute distance bug canvas 2.0 Future bug blaz new 2008-12-01T12:45:31+01:00 2008-12-01T12:45:31+01:00 "Ejla, ce dam wine dataset v attribute distance widget dobim tale exception: Unhandled exception of type AttributeError occured at 11:39:03: Traceback: File: OWBaseWidget.py, line 551 in processSignals Code: handler(value) File: OWAttributeDistance.py, line 103 in dataset Code: self.sendData() File: OWAttributeDistance.py, line 108 in sendData Code: matrix = self.computeMatrix() File: OWAttributeDistance.py, line 64 in computeMatrix Code: diss,labels = im.exportChi2Matrix() AttributeError: InteractionMatrix instance has no attribute 'exportChi2Matrix' " gregor Active Tickets 350 MDS nima legende canvas 2.0 Future bug somebody new 2009-03-02T18:30:29+01:00 2009-03-02T18:30:29+01:00 janez Active Tickets 373 poravnava dendrograma in barvne matrike v HeatMap canvas 2.0 Future bug somebody new 2009-03-25T06:58:05+01:00 2009-03-25T06:58:05+01:00 Pri kakrsnikoli vrednosti parametra Merge, ki je drugacna od 1, sta dendrogram in barvna matrika neporavnana. Primer: File->HeatMap, v File preberes iris.tab, v Filter das Merge na neko vrednost (npr. 15). blaz Active Tickets 379 Category name blank canvas 2.0 Future bug gregor new 2009-03-28T11:45:38+01:00 2009-03-28T11:45:38+01:00 When I add new category (Bioinformatics) through Canvas Options -> Widget tab order -> Add, widgets are added successfully but the Category name is not shown (blank). anonymous Active Tickets 437 napaka v izpisu Bayesovega klasifikatorja orng_modules 2.0 Future bug martin new 2009-05-05T14:26:20+02:00 2009-05-05T14:26:20+02:00 V modulu orngBayes se v funkciji printModel zatakne pri izpisovanju pogojnih porazdelitev za zvezne atribute. Glede na to, da gre za izpis modela, si sam ne upam popravljati / diskretizirati. lan Active Tickets 441 GEO Datasets- missing values canvas 2.0 Future bug somebody new 2009-05-07T10:46:28+02:00 2009-05-07T10:46:28+02:00 "- GEO Datasets widget ne dela pravilno, ce izbiras primere iz dolocenih razredov (izbere vse primere in ostalim kot razred doloci None) - fino bi bilo dodati se kaksno opcijo za missing values (remove genes with MV, remove genes with more than x% MV, impute MV)" Minca Active Tickets 450 Tooltip na scatterplotu/linear projection izgine canvas 2.0 Future bug somebody new 2009-05-26T15:45:22+02:00 2009-05-26T15:45:22+02:00 Ko premakneš miš ... ali pa včasih celo takoj, še preden jo premakneš. janez Active Tickets 485 Manjkajoca stran/datoteka: orngTextCorpus.htm web 2.0 Future bug blaz new 2009-07-29T21:43:14+02:00 2009-07-29T21:43:14+02:00 Link nanjo je na http://www.ailab.si/orange/doc/modules/. janez Active Tickets 486 Manjkajoca stran/datoteka: scree.png in biplot.png web 2.0 Future bug blaz new 2009-07-29T21:45:27+02:00 2009-07-29T21:45:27+02:00 "Sliki sta vkljuceni v stran http://www.ailab.si/orange/doc/modules/orngCA.htm " janez Active Tickets 487 Manjkajoca stran/datoteka: http://www.ailab.si/orange/doc/ofb/assoc3.py web 2.0 Future bug blaz new 2009-07-29T22:08:36+02:00 2009-07-29T22:08:36+02:00 "Nanjo kaže http://www.ailab.si/orange/doc/ofb/assoc.htm " janez Active Tickets 488 Manjkajoca stran/datoteka: http://www.ailab.si/orange/doc/reference/testRulelearner.py web 2.0 Future bug martin new 2009-07-29T22:24:02+02:00 2009-07-29T22:24:02+02:00 "Link nanjo je na strani http://www.ailab.si/orange/doc/reference/RuleLearner.htm To je tvoje, Martin, ne?" janez Active Tickets 491 "Manjkajoča stran/datoteka: """"" web 2.0 Future bug blaz new 2009-08-10T12:31:48+02:00 2009-08-10T12:31:48+02:00 "Tole je na doc/widgets/settings.htm: You can now also inspect the complete code of this widget. " janez Active Tickets 509 Nomogram crkne, če ima kak zvezen atribut samo eno vrednost canvas 2.0 Future bug martin new 2009-10-01T23:13:58+02:00 2009-10-01T23:22:29+02:00 Primer podatkov je v priponki. janez Active Tickets 549 Bug v picklanju SVM library 2.0 Future bug janez new 2010-01-15T23:22:54+01:00 2010-01-15T23:27:28+01:00 janez Active Tickets 589 orngStat.IS in BayesLearner z Laplace estimatorjem orng_modules 2.0 Future bug ales new 2010-05-14T14:36:41+02:00 2010-05-14T14:36:41+02:00 Na podatkih s konstantnim razredom ales Active Tickets 601 Ko widgti meljejo podatke bi se morale GUI kontrole disablat canvas 2.0 Future bug ales new 2010-05-25T11:42:23+02:00 2010-05-25T11:42:23+02:00 npr. v Test Learners dam leave one out, nato spremenim na cross-validataion se preden se konca. widget izracuna cross-validation nato pa nadaljuje z leave one out. ales Active Tickets 608 Spet problem z #RNGE pri unpicklanju library 2.0 Future bug janez new 2010-06-03T16:27:09+02:00 2010-06-03T16:27:09+02:00 Glej tu: http://www.ailab.si/orange/forum/viewtopic.php?p=2532#2532 janez Active Tickets 634 Schema/widget/global settings canvas 2.0 Future bug ales new 2010-08-04T14:13:05+02:00 2011-08-25T14:42:26+02:00 Vsi widgeti naj imajo lokalne/schema settinge ki se ne propagirajo visje razen ce se spremeniji (Ce sedaj odprem neko schemo, se vsi globalni settingi izgubujo) ales Active Tickets 652 Oklepaji v imenih signalov canvas 2.0 Future bug ales new 2010-11-18T13:08:32+01:00 2010-11-18T13:08:32+01:00 Canvas ne more naloziti widgetov, ki imajo v imenih signalov oklepaje ales Active Tickets 654 Passwords stored in schema files canvas 2.0 Future bug ales new 2010-11-19T20:13:23+01:00 2011-09-08T16:03:10+02:00 Usernames and passwords get stored in .ows files in plain text (i.e. PIPA widget). ales Active Tickets 655 Distribution.normalize naj preveri ce je distribucija res prazna library 2.0 Future bug janez new 2010-11-22T12:47:44+01:00 2010-11-22T12:47:44+01:00 "preden jo prepise {{{ >>> dist = orange.DiscDistribution(data.domain.classVar) >>> dist[0] += 20 >>> print dist, dist.abs <20.000, 0.000, 0.000>, 0.0 >>> dist.normalize() >>> print dist <0.333, 0.333, 0.333> }}} Dobro bi bilo tudi ce bi bila 'dist.abs' metoda (morda property). " ales Active Tickets 660 Napaka v pickleu library 2.0 Future bug janez new 2010-12-03T12:52:28+01:00 2010-12-03T12:52:28+01:00 Glej priloženi datoteki janez Active Tickets 662 orange.GraphAsList crash library 2.0 Future bug janez new 2010-12-09T12:52:17+01:00 2010-12-09T12:52:17+01:00 "naslednja koda: {{{ graph = orange.GraphAsList(2, False, 2) graph[0,1, 0] = 1 graph[0,1, 1] = 2 list(graph[0,1]) }}} povzroci segmentation fault" ales Active Tickets 691 BasketFeeder incorrectly respects locale library 2.0 Future bug janez new 2011-02-21T11:12:37+01:00 2011-02-21T11:12:37+01:00 "In Orange Canvas, reading a '.basket' file with specified decimal quantities (eg. ""reading=3.000"") on a machine with decimal separator set to something else than a period fails. TBasketFeeder::addItem (basket.cpp) uses strtod, which respects locale. Saving an Orange.data.Table to '.basket' file stores quantities with 3 decimal places using the dot separator, so Orange cannot even read files it produced. For example: import Orange Orange.data.Table('inquisition.basket').save('/tmp/test.basket') Now, the newly produced basket file cannot be read in Orange Canvas on a Linux machine with sl_SI.utf8 (Slovene UTF8) locale; File widget fails with ""'orange.BasketFeeder': invalid number after 'surprise=' in line 2"". In vars.cpp, TFloatVariable::str2val_low handles decimal separators correctly." matija Active Tickets 699 kNN with k bigger then the number of examples orng_modules 2.0 Future bug new 2011-03-10T15:57:31+01:00 2011-03-10T15:57:31+01:00 "When I set k larger then the numer of examples the AUC should be the same. {{{ import orange, orngTest, orngStat data = orange.ExampleTable('lenses.tab') res = orngTest.crossValidation([orange.kNNLearner(k=50), orange.kNNLearner(k=500)], data) print 'Number of examples:', len(data) print 'AUC:', orngStat.AUC(res) }}} output: {{{ Number of examples: 24 AUC: [0.8129032258064517, 0.31612903225806455] }}}" jurezb Active Tickets 705 Shortcut on Windows canvas 2.0 Future bug janez new 2011-03-17T15:31:09+01:00 2011-03-17T15:31:09+01:00 Shortcut to Orange should include python, e.g. c:\python26\pythonw to ensure that Orange starts on machines which have both 2.X and 3.X installed. janez Active Tickets 707 Rewrite the buildC4.5.py script library 2.0 Future bug ales new 2011-03-22T14:07:00+01:00 2011-03-22T14:07:00+01:00 using the distutils compiler class (Respect the standard ENV flags) ales Active Tickets 709 "ubuntu 10.10 ""no module named orngEnviron""" canvas 2.0 Future bug new 2011-03-25T01:13:12+01:00 2011-03-25T11:40:02+01:00 "Hi, I installed Orange on ubuntu 10.10 using sudo apt-get install. when I try to run orange by issuing the following line in a terminal, I get the message below: $ /usr/li/usr/bin/orange Traceback (most recent call last): File ""/usr/lib/python2.5/site-packages/orange/OrangeCanvas/orngCanvas.pyw"", line 7, in import orngEnviron, orngAddOns ImportError: No module named orngEnviron There is a file called orngEnviron.py in folder /usr/lib/python2.5/site-packages/orange/ with the following output from ls -l: -rw-r--r-- 1 root root 195 2011-03-23 22:21 orngEnsemble.py " rcrath Active Tickets 710 Change in orngText.py, Text addon canvas 2.0 Future bug crt assigned 2011-03-25T12:02:30+01:00 2011-03-28T14:36:21+02:00 "I think there is a missing check in TextClassifier in line 1646: ""if self.p_cond[c].has_key(w):"". Later I got snapshot of orange orange-source-snapshot-2010-06-30.zip and addons26.zip (attached). In that zip orngText has that check." szitnik Active Tickets 711 Vizrank crashes FreeViz canvas 2.0 Future bug new 2011-03-25T12:33:47+01:00 2011-03-25T12:33:47+01:00 Selecting projections in Vizrank while running freeviz crashes the latter. Freeviz should prevent changing the attribute list. janez Active Tickets 712 Genomics widget should save the selected organism names by string value canvas 2.0 Future bug ales new 2011-03-29T11:45:03+02:00 2011-03-29T11:45:03+02:00 not by integer index. After downloading of a server file for a new organism the index can be invalid because the organisms list changes. ales Active Tickets 720 ExampleTable.save() for .txt files doesn't preserve attribute metadata canvas 2.0 Future bug new 2011-04-02T00:18:46+02:00 2011-04-02T00:18:46+02:00 "The save() method doesn't preserve the special characters describing the variable type when writing new-style tab-delimited files (.txt). The special characters are decribed here: http://orange.biolab.si/doc/reference/tabdelimited.htm ""Prefixed attributes contain one- or two-lettered prefix, followed by ""#"" and the name. The first letter of the prefix can be either ""m"" for meta-attributes, ""i"" to ignore the attribute, and ""c"" to define the class attribute. As always, only one attribute can be a class attribue. The second letter denotes the attribute type, ""D"" for discrete, ""C"" for continuous, ""S"" for string attributes and ""B"" for baskets."" I know that at least the attribute type character is not being written, so saving then loading a .txt will not produce the same ExampleTable. Here is some code I've written to get around this (from http://orange.biolab.si/forum/viewtopic.php?f=3&t=1231): {{{ def savePar(path, table): with open(path, 'w') as f: # The order happens to match the order of the attributes in saveTxt attrs = ('%s%s#%s' % ( 'c' if a is table.domain.classVar else '', 'D' if a.varType == orange.VarTypes.Discrete else 'C', a.name) for a in table.domain) metas = ('mD#%s' % m.name for m in table.domain.getmetas().values()) print >> f, '\t'.join(itertools.chain(attrs, metas)) table.save(path + '.txt') with open(path, 'a') as f: with open(path + '.txt') as g: g.readline() f.write(g.read()) }}}" yang Active Tickets 722 Widget Purge domain removes meta attributes canvas 2.0 Future bug janez new 2011-04-06T13:22:38+02:00 2011-04-06T13:22:38+02:00 It does so even if when they are not constant. It shouldn't. janez Active Tickets 728 Scatter plot crashes on Show probabilities canvas 2.0 Future bug ales new 2011-04-07T16:15:35+02:00 2011-04-07T16:15:35+02:00 "I don't know how to reproduce this, but it happened as I connected File - Select Attributes - Scatter plot and checked Show probabilities. I haven't changed anything in Select attributes. After some clicking, the widget starts working. Apparently, xmax appears at some point... Unhandled exception of type AttributeError occured at 16:12:28: Traceback: File: OWScatterPlotGraph.py, line 509 in drawCanvas Code: self.computePotentials() File: OWScatterPlotGraph.py, line 475 in computePotentials Code: rx = self.transform(QwtPlot.xBottom, self.xmax) - self.transform(QwtPlot.xBottom, self.xmin) AttributeError: 'OWScatterPlotGraph' object has no attribute 'xmax' " janez Active Tickets 729 Purge Domain crashes due to camel case method name library 2.0 Future bug janez new 2011-04-07T16:26:29+02:00 2011-04-07T16:26:29+02:00 Fix the core, but then also change the widget. janez Active Tickets 731 saving Variable.attributes to .tab file library 2.0 Future bug janez new 2011-04-11T14:09:08+02:00 2011-04-11T14:09:08+02:00 "It does not work correctly when ""="" is in the string to be saved. For example: {{{ import Orange data = Orange.data.Table(""titanic.tab"") data.domain[0].attributes[""vceraj""] = ""bil lep=dan"" print data.domain[0].attributes data.save(""t.tab"") data = Orange.data.Table(""t.tab"") print data.domain[0].attributes }}} Prints: {{{ {'vceraj': 'bil lep=dan'} {'lep': 'dan', 'vceraj': 'bil'} }}}" marko Active Tickets 797 Fix the fake renaming of C++ classes other 2.0 Future bug janez new 2011-05-19T17:51:39+02:00 2011-05-19T17:51:39+02:00 "Aleš disabled it due problems with unpickling (mismatching ob_types). It would seem that only classes which are imported directly into new modules, without redefining them, may be renamed. This also makes sense from the user perspective - to actually construct an instance of the core class for classes which are redefined in modules, the user must use the core module, hence renaming would be misleading. In addition, Ales says: Da, problematični so tisti razredi ki so redefinirani (ni nujno da so izpeljani). Mislim, da je težava tudi omejena samo na tiste razrede ki vračajo 'ob_type' kot reconstructor v __reduce__ metodi (razredi z __pickleLoader-jem bi morali biti varni). " janez Active Tickets 800 import orngStat causes Segmentation fault and the system breaks orng_modules 2.0 Future bug new 2011-05-25T23:12:36+02:00 2011-09-10T17:26:05+02:00 "I have installed orange according to http://orange.biolab.si/svn/orange/trunk/orange/doc/INSTALL.linux.txt However, when I import orngStat, it causes Segmentaion error and the system broke out >>>import orngStat Segmentation fault demo@linux:$ My machine python 2.6.4 Ubuntu 64 bit 10.04 Does this have something to do with 64bit system? Thanks " nwisitti_og Active Tickets 807 Graphviz dot is not found on Mac OSX if not installed system-wide (in :/usr/bin:/usr/sbin:/sbin) other 2.0 Future bug ales new 2011-06-01T11:58:35+02:00 2011-06-01T11:58:35+02:00 when running the Orange.app bundle. Is it launched by LaunchServices and bypasses .bash_profile etc. ales Active Tickets 810 DomainContingency should ignore attributes which are not Discrete or Continuous library 2.0 Future bug janez new 2011-06-11T16:37:56+02:00 2011-06-11T16:37:56+02:00 Widget Distribution fails if the domain includes a string attribute. The problem lies in TDomainContingency::computeMatrix, which should not contruct TContingency for such attributes but put NULL there instead. As this is fixed, we should check for possible code which expects all elements of DomainContingency to be non-NULL, for instance TDomainContingency::normalize. janez Active Tickets 829 gsea orng_modules 2.0 Future bug marko assigned 2011-06-27T16:24:54+02:00 2011-07-01T14:36:40+02:00 Gsea widget: no warning if there are is no gene attribute or group attribute marko Active Tickets 867 Broken link in tutorial other 2.0 Future bug blaz assigned 2011-07-23T21:57:35+02:00 2011-07-25T12:02:06+02:00 "When following the ""Beginning with orange"" tutorial, the link to ""assoc.py is broken." CBrauer Active Tickets 904 Animation in Distributions Qt canvas 2.0 Future bug Noughmad new 2011-08-17T17:36:57+02:00 2011-08-22T17:46:42+02:00 I think there is no point in animating a probability point on variable change. However, the animation looks nice when the target variable is changed. Animation on enabling confidence intervals would only look nice if the interval would expand and move with the point. miha Active Tickets 920 C++ chrashes canvas 2.0 Future bug Noughmad new 2011-08-21T21:44:53+02:00 2011-09-26T13:52:21+02:00 Sometimes the C++ code chrashes. I'm not sure what happens and I can't replicate it. Think of a way to catch this nasty bug ;) miha Active Tickets 924 Different data sets can have the same checksum library 2.0 Future bug janez new 2011-08-22T10:37:13+02:00 2011-10-11T12:57:46+02:00 "The documentation data sets 'zoo.tab' and 'breast-cancer-wisconsin-cont.tab' have the same checksum, even though they contain completely different data. In visualization widgets (for example OWScatterplot) and in the 'preprocess.scaling' module, there are many checks to see if the new data is different from the old, and do an update only in this case. However, all those checks compare the checksums instead of the actual data sets. So when changing from 'zoo' to 'breast-cancer-wisconsin-cont', the plot data is not updated. The solution would be to ensure that checksums are unique, or to replace all test of like 'old_data.checksum() == new_data.checksum()' with 'old_data.checksum() == new_data.checksum() and old_data == new_data'. It would probably also be possible to override the __eq__ operator of ExampleTable so it would first compare the checksum, and then compare the data only if the checksums are the same. " Noughmad Active Tickets 929 Parellel Optimization does nothing canvas 2.0 Future bug janez new 2011-08-22T17:49:29+02:00 2011-08-22T19:59:57+02:00 The Parallel Optimization dialog doesn't seem to do anything. The behavior is common to both ParallelCoordinates and ParellelCoordinatesQt widgets, so I don't think it's related to visualization. Noughmad Active Tickets 937 Possible bug in handling missing attribute-values in MeasureAttribute other 2.0 Future bug janez accepted 2011-09-04T02:52:12+02:00 2011-10-03T17:34:34+02:00 "I came across this issue when attempting to construct regression trees (using TreeLearner) the simple way (without weights, pruning, etc.). Some of the attributes have missing values, and are getting up-weighted in the tree instead of being down-weighted by ReduceByUnknowns. In particular, I'm referring to code here: http://orange.biolab.si/trac/browser/trunk/source/orange/measures.cpp#L1052 I actually inserted log statements and discovered that outer.unknowns was 0 when the attribute-value being evaluated for split had 149 missing values out of a total of 179 examples (20 missing). Hence the MSE was unchanged. outer.cases=40; outer.unknowns=0 Thanks!" nkarthiks Active Tickets 949 Debian package: convert copyright file to DEP5 other 2.0 Future bug new 2011-09-11T11:32:46+02:00 2011-09-11T11:32:46+02:00 We should convert `copyright` file to [http://dep.debian.net/deps/dep5/ DEP5]. mitar Active Tickets 963 vizrank: number of instances require a minimum setting other 2.0 Future bug new 2011-09-15T12:16:59+02:00 2011-09-15T12:16:59+02:00 "Dear all! I frequently use visrank to assemble marker combinations. However the algorithm has its trouble with data with lots of NA (non availables). The problem is that in a combination of variables the number of #instances gets lower and lower, while the classification accuracy seems to be good (no wonder with few instances !!) Is ist possble to implement under ""settings"" a field that the number of instances shall be larger then a value to be selected? #instances > 10 for example This would even save time in the vizrank evaulation loop rejecting non relevant marker combinations because they have insufficient instances. " lareooreal Active Tickets 977 Erroneous references to Orange.statistics.Distribution in documentation other 2.0 Future bug marko new 2011-10-06T16:34:56+02:00 2011-10-06T16:34:56+02:00 "There are many references to `Orange.statistics.Distribution`, which does not exist, in the documentation. I am not sure whether that's a result of copying and pasting an erroneous reference, or the class was refactored later. Probably the references should be corrected to `Orange.statistics.distribution.Distribution`. An alternative is to import relevant classes higher up in the hierarchy (like `Orange.network` imports `Graph` from `Orange.network.network`, but that's a completely different situation) to make `Orange.statistics.Distribution` a correct (and probably preferred) name of the class." matija Active Tickets 980 We should probably depend on python-qt4-gl on Debiana and Ubuntu systems other 2.0 Future bug miha assigned 2011-10-07T22:07:32+02:00 2012-05-29T11:01:30+02:00 VisualizeQt widgets don't work on Ubuntu 11.10, unless the python-qt4-gl package is installed. I believe this is also necessary (and exists) on older Ubuntu and Debian systems, but I haven't checked it. We should probably add a dependency, because there was already a [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1393 complaint on the forum] because of this. matija Active Tickets 1014 Discretizer missing in windows install? canvas 2.0 Future bug new 2011-11-30T13:20:20+01:00 2011-11-30T14:41:03+01:00 "Hi, I am trying to use the software for a data mining applications course. Some students download and install the package and they have the discretizer widget, others download and they do not have the widget. It seems to depend on whether the package is downloaded from the main page or the download page. Any ideas why this is happening, or suggestions for which is the correct package to download to get the full set of supported widgets? Thanks" dgroth Active Tickets 1032 netexplorer: zoom (when you click the zoom icon) does not work canvas 2.0 Future bug marko assigned 2011-12-14T10:36:33+01:00 2011-12-20T22:48:26+01:00 Zomming with control + mouse works, but the icon click does nothing. Also applies to Scatterplot (Qt) and possibly other Qt widgets. marko Active Tickets 1045 debian - control file is not update, numpy is missing in the dependencies other 2.0 Future bug mitar accepted 2011-12-20T14:48:57+01:00 2012-05-29T11:32:50+02:00 . marko Active Tickets 1050 packaging: default python for PyQt4-Qwt5 and networkx other 2.0 Future bug ales new 2011-12-23T14:17:39+01:00 2011-12-23T14:17:39+01:00 "Installers for PyQt4-Qwt5 and networkx offer Python 3.2 by default, if it is installed. A user that just keeps clicking ""Next"" therefore installs them to the wrong Python." marko Active Tickets 1053 Segfault when using Sphereviz 3D. canvas 2.0 Future bug matejd new 2012-01-09T12:30:35+01:00 2012-01-09T12:30:35+01:00 See this forum thread [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1444] ales Active Tickets 1055 SQL module does not recognize discrete attributes other 2.0 Future bug new 2012-01-11T23:42:03+01:00 2012-01-11T23:42:03+01:00 "Dear Janez; I have changed and add D# as the prefix of the column name, but it still doesn't work Yes, this OWSQL.py never out form the prototype. I believe that this module is very useful for Orange as a data mining tool. Your help is appreciated regards, invanos On 1/8/2012 4:58 PM, Janez Demsar wrote: > Hi, > > the SQL widget never made it out from prototypes, I guess. But it > should be handle discrete attributes (I used it quite a lot some time > ago). Can you try what happens if you change the column name to > D#rkn_id - as I remember, the prefix D# can be used to explicitly tell > Orange to treat the attribute as discrete. > > Bye, > Janez > > On Sun, Jan 8, 2012 at 10:32 AM, invanos wrote: >> Hi Janez; >> >> I am new in Orange Canvas, and very amazed with this software. >> >> I am planning to use Orange as part of my big project that will be >> connected to Postgre & MySQL. >> I have been trying to explore the possibility to use OWSQL module as the >> entry point to access the above database. >> It gives me error if the field contains text, but success for numeric. >> >> The error is "" >> Error: 'orange.FloatVariable': 'a' is not a legal value for continues >> attribute 'rkn_id' >> >> Note --> a is the content of rkn_id field, which is text >> >> my question is >> a. Is OWSQL only able to access numeric values? >> b. How to handle discrete variables? I assume that it always treats >> numeric as continues variable? >> c. How to resolve the above case, if the table contain text? >> >> Your assistance and respond is highly appreciated >> >> Thank you >> >> regards, >> Invanos >>" janez Active Tickets 1093 Lookup classifier library 2.0 Future bug janez new 2012-02-08T16:27:13+01:00 2012-02-08T16:27:13+01:00 "According to the [http://orange.biolab.si/doc/reference/Orange.classification.lookup/ documentation], the output of the appended script should be {{{ ['1', '1', '1', '1'] ['1', '1', '2', '1'] ['1', '1', '3', '1'] ['1', '1', '4', '1'] ['1', '2', '1', '1'] ['1', '2', '2', '0'] ['1', '2', '3', '0'] ['1', '2', '4', '0'] ['1', '3', '1', '1'] ['1', '3', '2', '0'] }}} but the script actually returns {{{ ['1', '1', '1', '?'] ['1', '1', '2', '?'] ['1', '1', '3', '?'] ['1', '1', '4', '?'] ['1', '2', '1', '?'] ['1', '2', '2', '?'] ['1', '2', '3', '?'] ['1', '2', '4', '?'] ['1', '3', '1', '?'] ['1', '3', '2', '?'] }}} " jurezb Active Tickets 1107 GO Browser widget (and obiGO) must allow specifing the exact annotation/ontology release addon/bioinformatics 2.0 Future bug ales new 2012-02-23T10:54:39+01:00 2012-02-23T10:54:39+01:00 to use. ales Active Tickets 1185 pick_class #RNGE bug library Future bug janez assigned 2012-05-08T19:46:29+02:00 2012-05-31T12:42:08+02:00 "Sometimes when working with multi-target data and using the pick_class function the remaining classes get corrupted with values outside their range, represented by #RNGE. This is a sample of code that reproduces the error. In the first one the error is minor and only produces two #RNGEs, the second example that also changes the domain to include the picked class into the features shows the escalation of the problem. d = Orange.data.Table('bridges.v2.nm') for i in [3,2,1,0]: i = random.randrange(0,3) print d.domain.class_vars[i] d.pick_class(d.domain.class_vars[i]) for x in d: print x d = Orange.data.Table('bridges.v2.nm') for i in [3,2,1,0]: print d.domain.class_vars[i] d.pick_class(d.domain.class_vars[i]) for x in d: print x d.change_domain(Orange.data.Domain(d.domain, False, \ class_vars=d.domain.class_vars))" mlevar Active Tickets 1265 Prototypes/Transpose widget fails with unknown values canvas Future bug ales new 2013-01-09T11:13:54+01:00 2013-01-09T11:13:54+01:00 Load brown-selected dataset and connect to the widget ales Active Tickets 1268 'Save as matplotlib script' does not always work. canvas Future bug ales new 2013-01-21T10:51:16+01:00 2013-02-15T15:37:14+01:00 In fact it rarely works. The option should probably be removed. A way to fix it would be to implement a [http://doc.qt.digia.com/qt/qpaintdevice.html QPaintDevice], [http://doc.qt.digia.com/qt/qpaintengine.html QPaintEngine] pair (similar to how QSvgGenerator works) ales Active Tickets 1274 "ScatterPlot ""Auto Send"" settings are in the wrong tab" canvas Future bug new 2013-01-31T17:16:01+01:00 2013-01-31T17:16:01+01:00 "The ""Auto Send ..."" check box is in a different tab then the ""Select"" controls. This is not a good design as it is not immediately clear why selection does not send the data if the auto send is turned off." ales Active Tickets 1280 Graphical glitch: the lines between widgets sometimes break up in two parts new-canvas Future bug new 2013-02-14T14:08:48+01:00 2013-02-14T14:08:48+01:00 When there is a connection between two widgets, where the second one (the receiving one) is placed to the right and up of the first widget, under some angles the link graphically breaks up in two in the middle. matija Active Tickets 1281 "Too narrow widget ""ears"" for links" new-canvas Future bug new 2013-02-14T14:15:00+01:00 2013-02-14T14:15:00+01:00 "While visually it looks OK, it is hard to position the mouse over widget's ""ears"" to create a link. The sensitive area should be, in my opinion, extended by a few pixels in each direction." matija Active Tickets 1307 Modal dialogs do not have a close window button in GNOME 3 new-canvas Future bug ales new 2013-05-24T12:56:30+02:00 2013-05-24T12:56:30+02:00 This makes the Welcome Screen and Settings somewhat difficult to dismiss (i.e by pressing ESC, but this is not obvious). ales Active Tickets 1309 Update Python and PyQt4 in the windows installer packages. other Future bug ales new 2013-05-29T18:04:08+02:00 2013-05-29T18:04:08+02:00 The current installer for Orange for Python 2.6 contains an old version of PyQt4 which is not even able to start the new interface (Orange 2.7) ales Active Tickets 790 continuize widget segmentation fault? canvas 2.0 Future check new 2011-05-06T11:48:20+02:00 2011-05-06T13:43:14+02:00 "We should test this. [11:40:25 AM] x: I think I found solution [11:40:38 AM] x: everything was just fine [11:41:10 AM] x: and suddenly continuize widget started to crash [11:41:37 AM] x: i thought it was something related to mac platform but it started with my pc too [11:42:18 AM] x: i was using some of predefined data sets that comes with orange [11:46:09 AM] x: of course I added data input, but it crashes when i click on continuize, anyway i think i solved the problem by cleaning widget settings and reseting orange, it is ok for now... thanks " marko Active Tickets 885 bug in bestThreshold library 2.0 Future check janez new 2011-08-08T14:46:43+02:00 2011-08-08T14:46:43+02:00 "bestThreshold method of Orange.feature.scoring.Measure (previously, orange.MeasureAttribute) was returning the score instead of the threshold. Was fixed in measures.cpp (http://orange.biolab.si/trac/changeset/11452), but has to be rechecked - also elsewhere." lanz Active Tickets 951 Check if we can remove -lstdc++ when compiling other 2.0 Future check new 2011-09-11T16:19:32+02:00 2011-09-11T16:19:32+02:00 Check if we can remove -lstdc++ when compiling, as it probably not necessary if using g++. mitar Active Tickets 952 Check if we can build Orange libraries dynamically-only on Linux other 2.0 Future check new 2011-09-11T16:21:55+02:00 2011-09-11T16:21:55+02:00 "Currently we are building static liborange_include.a and then continue building+linking shared libraries. We should check if this is really necessary to do so or maybe we can build them all shared. Maybe we just didn't know how to do it properly at that time. Or maybe there are some other problems (references, counting) why this is necessary so. It is not necessary on Windows." mitar Active Tickets 954 Debian package: run lintian and correct errors other 2.0 Future check new 2011-09-11T18:59:40+02:00 2011-09-11T18:59:40+02:00 lintian is a tool which checks packages for conformity to the Debian Packaging Policy. We should run it on our package and correct any errors we might find. mitar Active Tickets 1016 Regression test fails orng_modules 2.0 Future check marko new 2011-12-01T12:53:13+01:00 2011-12-01T12:53:13+01:00 "Please check and fix the following tests: abcn2 and other rules, fss, imputation, k-means, ensembles (this is probably correlated with the tree3.py test failure). Test results: http://orange.biolab.si/download/regressionLogs/winxp/regression_tests_orange_report.html" miha Active Tickets 1017 Regression test fails orng_modules 2.0 Future check anze new 2011-12-01T12:54:34+01:00 2011-12-01T17:01:48+01:00 "Regression test fails Please check and fix the following tests: PCA, bayes and imputation. Test results: http://orange.biolab.si/download/regressionLogs/winxp/regression_tests_orange_report.html" miha Active Tickets 1279 """Scheme"" vs. ""Schema""" new-canvas Future check new 2013-02-14T14:03:52+01:00 2013-02-14T14:03:52+01:00 "I disagree with the usage of word ""scheme"" in place of ""schema"" that has been used in the old canvas. ""Scheme"" sounds more like a ""plan, where ""schema"" sounds more like a ""diagram"", which it really is. I propose some kind of voting or an argumented debate in the comments." matija Active Tickets 203 Latch v canvasu canvas 2.0 Future task somebody new 2008-03-25T22:56:46+01:00 2009-03-11T12:46:08+01:00 janez Active Tickets 330 Feature list na webu web 2.0 Future task blaz new 2009-01-29T12:42:54+01:00 2009-01-29T12:42:54+01:00 janez Active Tickets 335 Screencasts web 2.0 Future task blaz new 2009-01-29T12:49:47+01:00 2009-06-22T17:41:10+02:00 "Pripraviti je potrebno malo besedila, potem pa se, najbrz, dogovoriti za format screencastov, tako da bi lahko screencaste kar odlagali v nek direktorij, pa bi se avtomatsko zlozili na stran. Za zacetek pa je potrebno sploh pripraviti nekaj screencastov..." janez Active Tickets 481 "Extensions v koloni ""Home""" canvas 2.0 Future task blaz new 2009-07-23T14:51:14+02:00 2009-07-23T14:51:14+02:00 janez Active Tickets 618 Kaj se manjka v oknu Report canvas 2.0 Future task janez assigned 2010-07-04T09:51:00+02:00 2010-11-07T23:54:31+01:00 "Gumb Clear in Print Tipka Delete, ko si v listboxu" janez Active Tickets 703 Attribute statistics widget should be cleaned and improved canvas 2.0 Future task new 2011-03-16T09:33:26+01:00 2011-03-16T09:35:34+01:00 "It currently includes two different visualizations - histograms and box plots. Histograms are already provided in Distributions. Is a separate widget really needed? Box plots should allow for - comparison between distributions by drawing multiple box plots side by side - allow horizontal layout for comparison of greater number of box plots - lines representing the density should not be centered but left- (or bottom-) aligned" janez Active Tickets 798 Which learning algorithms can handle unknown values? other 2.0 Future task new 2011-05-19T22:07:55+02:00 2011-05-19T22:07:55+02:00 Somebody should verify the documentation about learning algorithms; each should explicitly state what happens with unknown values. janez Active Tickets 811 Smart legend for OWGraph canvas 2.0 Future task Noughmad accepted 2011-06-12T13:37:02+02:00 2011-08-22T14:15:03+02:00 "The OWGraph class should know how its curves are classified (by size, shape and color), and should produce corresponding legend entries. When data points are classified by at least two attributes (let's say color and size), the legend should consist of two tables: one for color and one for size. It should also be noted which data attributes these two variables represent. Currently, a similar system is implemented inside OWScatterPlot, but it should be done inside OWGraph and Legend classes, so that graph classes don't have to bother with it. " Noughmad Active Tickets 830 gsea: select whole columns orng_modules 2.0 Future task marko assigned 2011-06-27T16:26:05+02:00 2011-07-01T14:37:00+02:00 If there are no group specification, should we allow users to make groups of phenotypes themselves? marko Active Tickets 845 Calibration plot canvas 2.0 Future task new 2011-07-01T09:14:35+02:00 2011-07-01T09:14:35+02:00 The widget was neglected for long and needs to be checked and partially rewritten (Qwt -> Qt etc). janez Active Tickets 848 Debian packages for add-ons other 2.0 Future task new 2011-07-03T12:54:58+02:00 2011-07-03T12:54:58+02:00 Create Debian packages also for add-ons. mitar Active Tickets 854 Orange.projection.correspondence documentation 2.0 Future task assigned 2011-07-04T11:12:40+02:00 2012-02-14T11:29:43+01:00 That was assigned to you, wasn't it? marko Active Tickets 878 Unit tests for the 'plot' module canvas 2.0 Future task Noughmad new 2011-08-02T17:12:06+02:00 2011-08-22T17:43:54+02:00 "The new plot module has no unit tests, add some. At least point selection management should have a set of tests. " Noughmad Active Tickets 892 Plots in VizRank dialog other 2.0 Future task new 2011-08-16T15:17:58+02:00 2011-08-16T15:17:58+02:00 "There are for simple plots in VizRank that show projection quality and interactions between attributes. They are all very simple, but they don't work with OWPlot yet. Make them use OWPlot. " Noughmad Active Tickets 907 Add some more things to tunk canvas 2.0 Future task Noughmad new 2011-08-17T17:50:18+02:00 2011-08-17T17:50:18+02:00 Update trunk to support orngScaleLinProjData3D. See orngScaleLinProjData.py. miha Active Tickets 916 Buttons in toolbar canvas 2.0 Future task Noughmad accepted 2011-08-21T21:31:21+02:00 2011-08-22T10:54:41+02:00 "Remove the pan button. Add pan to the right + hold in zoom mode. Remove the three buttons for selection type. Add an overlay menu that appears when you click + hold the select mode button. The three buttons should appear one after another above the select mode button when toolbar is displayed horizontally and right of the select mode button when toolbar is displayed vertically. User should release the button over the desired selection type. When the user releases the button, the overlay menu should hide." miha Active Tickets 947 Debian package: move documentation for canvas and library into separate packages other 2.0 Future task new 2011-09-11T11:25:28+02:00 2011-09-11T11:27:30+02:00 "Move documentation for canvas and library into separate Debian packages. Maybe also a package for datasets? There are some datasets also among documentation (not in `datasets` subdir)." mitar Active Tickets 948 Debian package: honor DEB_BUILD_OPTIONS other 2.0 Future task new 2011-09-11T11:31:50+02:00 2011-09-11T11:31:50+02:00 "We should honor DEB_BUILD_OPTIONS when compiling Debian package. We can simply support that in our makefiles. [http://www.debian.org/doc/debian-policy/ch-source.html#s-debianrules-options Related documentation]." mitar Active Tickets 953 Debian package: convert embedded libraries to external dependencies library 2.0 Future task assigned 2011-09-11T18:55:15+02:00 2011-09-22T11:53:59+02:00 "For inclusion of our Debian package into official repository we have to move embedded libraries to external dependencies. Those are (at the time of writing): * [source:trunk/source/orangeom/qhull qhull] * [source:trunk/source/orange/blas blas] * [source:trunk/source/orange/linpack linpack] There are [http://packages.debian.org/squeeze/libqhull-dev libqhull-dev], [http://packages.debian.org/squeeze/libblas-dev libblas-dev] and [http://packages.debian.org/squeeze/liblapack-dev liblapack-dev], respectively. And probably libSVM (at least we have it mentioned in licenses file)." mitar Active Tickets 955 Debian package: change versioning other 2.0 Future task new 2011-09-11T19:01:57+02:00 2011-09-11T19:01:57+02:00 Currently we have packages versioned as `PACKAGENAME_0.0.REVISION~svn-1`, but it should be `PACKAGENAME_2.0+svnREVISION-1`. mitar Active Tickets 1022 Random Forest should use SimpleTreeLearner by default. other 2.0 Future task new 2011-12-05T15:36:54+01:00 2011-12-05T15:36:54+01:00 I added the argument random_generator to SimpleTreeLearner. It's used in the same way as in MakeRandomIndices. See the sample file attached. Can you make random forests use SimpleTreeLearner by default? Thanks. jurezb Active Tickets 1026 SimpleTreeClassifier tree attribute library 2.0 Future task jurezb new 2011-12-06T15:55:00+01:00 2011-12-06T15:55:00+01:00 SimpleTreeClassifier should have a tree attribute pointing to the root node. marko Active Tickets 1027 Add Python Qt OpenQL to requirements. other 2.0 Future task miha assigned 2011-12-09T15:08:11+01:00 2012-05-29T11:01:17+02:00 miha Active Tickets 1049 download links that are stable other 2.0 Future task miha assigned 2011-12-23T11:55:45+01:00 2012-05-29T11:00:39+02:00 "Our download links include dates and revisions, like http://orange.biolab.si/download/orange-source-snapshot-2011-12-21.zip. This is very good, but we also need some stable download links, that would always return the newest version. Something like http://orange.biolab.si/download/orange-source-snapshot.zip. " marko Active Tickets 1083 Check docstrings in documentation with doctest. documentation 2.0 Future task jurezb new 2012-02-05T22:13:12+01:00 2012-02-05T22:13:12+01:00 jurezb Active Tickets 10 Attribute Statistics: napisati na novo library Future wish new 2007-05-02T10:54:18+02:00 2008-08-05T16:56:58+02:00 Widget je napisan nekoliko nerodno, verjetno bi ga bilo mogoce bistveno poenostaviti. janez Active Tickets 13 Logisticna regresija: prikaz odds ratio library Future wish martin new 2007-05-02T10:57:47+02:00 2013-05-11T09:42:42+02:00 Potrebovali bi widget, ki pokaze odds ratio za logisticni regresijo. Za to je zdaj sicer mogoce uporabljati nomogram, vendar je to nerodno. janez Active Tickets 15 Load File wizard library Future wish new 2007-05-02T11:00:26+02:00 2008-08-04T20:28:25+02:00 "Zelo koristno bi bilo pripraviti wizard, ki bi ti pomagal, kadar se file ne nalozi. Glavni problem pri uporabi Orangea je priprava filea s podatki in napake so navadno trivialne, vendar uporabnika odbije, ker mu stvar ze v zacetku ne dela. Stvar bi naredili, za zaceteke, za .tab file. ""Orange seems to have a trouble with your data file. This wizard can help you find the problem."" ""cancel, next"" try: f = file(filename, ""rt"") except: """"""File cannot be opened for reading: it either doesn exist or another application (such as Excel, if you are using Windows, doesn allow opening it."""""" ""ok"" firstline = f.readline() while (not firstline or firstline[0]==""|"" or not firstline.trim()) and not f.eof(): firstline = f.readline() if f.eof(): """"""Your file seems to be empty or it only contains comments (lines starting with character |). Please open it in an ASCII editor and check this"""""" attr_names = firstline.split(""\t"") if len(attr_names) <= 1: """"""The first line should contain a list of attribute names, separated with tabulators. There are not tabulators in your first line, so it would appear to have only one attribute. If this is correct, press ""Next"". Otherwise, if you have prepared the file using a spreadsheet application, such as Excel, make sure that you have saved it as tab-delimited. If you have prepared it using an ASCII editor, verify whether you used tabulators and whether your editor keeps them (some editors can be set to replace tabulators with spaces). Do we need to go ahead?"""""" ""no, next"" """"""Your file has %i attributes (possibly including the class attribute), named as shown in the table below. Is this correct?"""""" ""no, yes"" no-> """"""If you miss some attributes or some attribute names are merged into a single attribute, there must be something wrong with the tabulators that should be separating them. If you have prepared the file using a spreadsheet application, such as Excel, make sure that you have saved it as tab-delimited. If you have prepared it using an ASCII editor, verify whether you used tabulators and whether your editor keeps them (some editors can be set to replace tabulators with spaces)."""""" secondline = f.readline() while (not secondline or secondline[0]==""|"" or not secondline.trim()) and not f.eof(): secondline = f.readline() if f.eof(): """"""Your file seems to contain nothing but the attribute names. Please open it in an ASCII editor and check this"""""" attr_types = secondline.split(""\t"") if len(attr_types)==1 and len(attr_names)==1 and secondline.count("" ""): """"""There are no tabulators in the second line, which gives the attribute types. You have more than one attribute, hence the number of attribute types mismatches the number of attributes. Is it possible that you have separated attributes types with spaces instead of with tabulators? Anyway, we have found the problem; fix this and try again."""""" ""ok"" if len(attr_types > len(attr_names): """"""There are more attribute types than attribute names. Please open the file in an ASCII editor or, even better, in a spreadsheet, such as Excel, and locate the problem"""""" if len(attr_types) < len(attr_names): """"""The number of attribute types (the second line) is smaller than the number of attributes (that is, the number of attribute names in the first line). The table shows which type went with each attribute; locate the source of mismatch and fix the corresponding entry in the file"""""" !now check the attribute types, put them in t y again."""""" """"""Orange was unfortunately unable to determine the source of your problem. Please consult the documentation on file format and the error message returned by the data loader (%s) to try to fix the problem yourself. If your data is not confidential you can also send us the file (or a part of it; include a small portion which still does not load), and we can try to make this wizard smarter.""""""" janez Active Tickets 16 Load File: splosni bralnik library 2.0 Future wish ales assigned 2007-05-02T11:01:29+02:00 2012-03-19T11:51:15+01:00 Zanimivo bi bilo imeti branje iz poljubnega tekstovnega filea, kjer uporabnik sam dolo?i delimiterje ipd. Podobno kot v Excelu in Knimeu janez Active Tickets 52 Widget, ki podatke posilja po casu Future wish new 2007-05-15T15:57:30+02:00 2007-05-15T15:57:30+02:00 "Naredili bi widget, ki bi mu povedali, kateri atribut predstavlja cas, potem pa bi posiljal ven eno za drugo ExampleTable, ki bi vsebovala vse primere z doloceno vrednostjo tega atributa. Widget bi imel gumb Play, pa tudi slider, za rocni play. Glej: http://video.google.com/videoplay?docid=2670820702819322251 http://tools.google.com/gapminder/" janez Active Tickets 135 textAttributePos to be determined automatically orng_modules 1.0 Future wish sasa new 2007-11-29T11:16:47+01:00 2008-05-19T01:31:34+02:00 "In orngText all the functions that require textAttributePos should have this with default to None, and if not given, should determine it to be the last string attribute in the data. Documentation and examples should be changed accordingly." blaz Active Tickets 140 OWCorrAnalysis canvas 1.0 Future wish sasa new 2007-11-30T09:26:12+01:00 2008-05-19T01:32:20+02:00 This widget should allow users to use similar interface for selection of point labels and colors as Scatterplot widget (Colors and Point labels in OWScatterplot's Main tab) blaz Active Tickets 165 Urejanje podatkov o stestiranih learnerjih v Test Learners widgetu canvas 1.0 Future wish blaz new 2007-12-31T14:57:07+01:00 2008-03-13T21:23:25+01:00 Če testiraš večje število learnerjev, bi bilo dobro, če bi lahko rezultate, če jih že imaš v tabelici, tudi uredil po katerem od stolpcev tam. mitar Active Tickets 518 v GEO dodaj !Series_submission_date = Jun 28 2006 orng_modules 2.0 Future wish blaz new 2009-11-04T12:57:09+01:00 2009-11-04T12:57:09+01:00 " {{{ - iz GDS dobis !dataset_reference_series = GSE5178 - gres na ftp://ftp.ncbi.nih.gov/pub/geo/DATA/SOFT/by_series/GSE5178/ - zloadas GSE5178_family.soft.gz - vzames !Series_submission_date = Jun 28 2006 in !Series_last_update_date = Sep 10 2007 - vzami se podatke o Series_contact }}} " blaz Active Tickets 546 Attribute Statistics - reusable visualization canvas 2.0 Future wish ales new 2010-01-13T13:44:48+01:00 2010-01-13T13:44:48+01:00 Attribute stat. visualization should be reusable (in ClassificationTreeGraph, RegressionTreeGraph, ...) ales Active Tickets 562 UNIX - Orange Canvas ustvari prazen direktorij ~/orange-reports/ canvas 2.0 Future wish somebody new 2010-02-23T21:24:27+01:00 2010-02-23T21:24:27+01:00 "Ob zagonu Orange Canvas se ustvari vsaj zaenkrat nepotreben prazen direktorij ""orange-reports"" v home direktoriju uporabnika. Pricakovano bi bilo, da se direktorij ustvari sele, ko ga program potrebuje. Trenutno je to (in za ostale operacijske sisteme) kar hardcodeano v ""orngEnviron.py"". Sistem: - Debian Squeeze (testing) - g++-4.4 4.4.2-9 - python2.6 2.6.4-5 - python-dev 2.5.4-9 - python-imaging 1.1.7-1+b1 - python-matplotlib 0.99.1.2-3 - python-numpy 1:1.3.0-3+b1 - python-numpy-ext 1:1.3.0-3 - python-qwt5-qt4 5.2.1~cvs20091107+dfsg-3 - orange (2.0b) in orange/source iz svn rev 8399" gw.orange@… Active Tickets 640 Show multiple component planes in SOM Visualizer widget canvas 2.0 Future wish ales reopened 2010-09-20T10:57:21+02:00 2013-01-21T11:38:14+01:00 ales Active Tickets 648 Izboljsati implementacijo orngStat.compute_CD() (critical difference) orng_modules 2.0 Future wish somebody new 2010-11-07T04:05:37+01:00 2010-11-07T17:42:02+01:00 "Funkcija `orngStat.compute_CD()` je implementirana bolj slabo (s hardcodeanimi ''q'' vrednostmi za teste) in se sesuje v kolikor zelis primerjati med seboj vec kot 10 metod. Pravilna resitev bi bila, da izracuna potrebno ''q'' vrednost za kar pa mi na internetu ne uspe najti algoritma ali literature s formulo. V kolikor je izracun dolgotrajen proces pa bi bilo boljse imeti se vedno hardcodeane vrednosti za nizje stevilo primerjanih metod, za visje pa sproti poracuna." gw.orange@… Active Tickets 682 Widget za learning curve canvas 2.0 Future wish somebody new 2011-01-07T00:20:52+01:00 2011-01-07T00:20:52+01:00 Kaže naj tako rezultate na testni kot tudi učni množici (iz razlike se da sklepati, ali imaš težave s pristranskostjo ali varianco...) janez Active Tickets 693 Implementacija algoritma ReliefFS orng_modules 2.0 Future wish somebody new 2011-02-28T22:55:39+01:00 2011-02-28T22:55:39+01:00 "Po clanku od ''Peer, Čargo and Kononenko (1997)'' sodec je obcutno boljsa, dokaj preprosta, a procesorsko zahtevnejsa izboljsava algoritma ReliefF imenovana ReliefFS. Gre za idejo poganjanja navadnega algoritma ReliefF za vse mozne vrednosti ''k''. Preprosta implementacija tega se najde v diplomski: [http://rudnik.tnode.com/hg/file/e88fc30465e1/src/MeasureAttribute_reliefFS.py]" gw.orange@… Active Tickets 696 Fix memory useage of Bioinformatics modules, widgets orng_modules 2.0 Future wish ales new 2011-03-09T16:09:44+01:00 2011-03-09T16:09:44+01:00 For instance: just loading the Human annotations in GO and pathways in KEGG widget takes nearly 1GB. ales Active Tickets 701 Ontology graph visualization widget. orng_modules 2.0 Future wish ales new 2011-03-11T10:54:52+01:00 2011-03-11T10:54:52+01:00 Implement an ontology graph visualization widget/framework (for any ontology from OBOFoundry (obiOntology module, see EBI - Ontology Lookup Service for a general idea). Also use it to (re)implement the OWGOGraphTermFinder from orange 1.0 (or use the framework to extend the GO Enrichment widget) ales Active Tickets 743 Merge the two fixers in the refactoring tool. orng_modules 2.0 Future wish ales new 2011-04-21T12:07:07+02:00 2011-04-21T12:07:07+02:00 The imports and class renames are part of the same complex restructuring (maybe split it into refactoring stages). ales Active Tickets 744 Stoppable widget processing. canvas 2.0 Future wish ales new 2011-04-21T12:23:17+02:00 2011-04-21T12:23:53+02:00 Actions in widgets should be stoppable if possible. The progressBarSet method could raise a special exception (subclass of KeyboardInterupt). This would be handled by a function decorator that would also update the gui (show the Stop button (in the widget as well as in the Canvas). ales Active Tickets 805 Create 64-bit installer for windows and dmg for OSX library 2.0 Future wish ales new 2011-05-26T21:16:14+02:00 2011-05-26T21:16:14+02:00 ales Active Tickets 812 A stable build on the front page. other 2.0 Future wish ales new 2011-06-13T10:26:00+02:00 2011-06-13T10:26:00+02:00 "As daehyok [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1322 suggests] we should be linking to a stable tested orange build from the front page. The nightly builds should still downloadable from the ""Nightly Builds"" page." ales Active Tickets 819 Finish the unittest framework other 2.0 Future wish ales new 2011-06-15T12:54:46+02:00 2011-06-15T12:54:46+02:00 and write comprehensive test cases. ales Active Tickets 923 Wish: Manual movement of attributes in Linear Projection (Qt) canvas 2.0 Future wish Noughmad new 2011-08-22T10:13:25+02:00 2011-08-22T10:13:25+02:00 It'd be nice if one could grab an attribute handle with a mouse and move it to an arbitrary location, thus altering the projection manually. matija Active Tickets 927 Add 'Use input matrix as Contingency matrix' option to Correspondence widget canvas 2.0 Future wish ales new 2011-08-22T15:17:37+02:00 2011-08-22T15:17:37+02:00 ales Active Tickets 931 Better csv support. other 2.0 Future wish ales new 2011-08-26T10:19:36+02:00 2011-09-08T15:27:47+02:00 Add read and write methods to io module for csv (using standard python csv module) replacing the built-in csv reader. ales Active Tickets 941 Finish array express data matrix (fgem) loading. other 2.0 Future wish ales new 2011-09-08T14:54:39+02:00 2011-09-08T14:54:39+02:00 ales Active Tickets 968 The Renderer class in Orange.misc.render should implement a format independant render logic orng_modules 2.0 Future wish ales new 2011-09-27T16:38:59+02:00 2011-09-27T19:01:20+02:00 And should be used in place of SVG, ESP, PIL renderers. Saving to a file should then be done by specialized serializers. ales Active Tickets 978 Merge data should also be able to merge data by index canvas 2.0 Future wish ales new 2011-10-07T12:06:03+02:00 2013-04-15T10:38:16+02:00 if both inputs are of the same length. ales Active Tickets 983 Add an Aditional data set input to Associate/Itemset Explorer/Filter canvas 2.0 Future wish ales new 2011-10-14T18:47:53+02:00 2011-10-14T18:47:53+02:00 "(or maybe a new widgets Apply Rules/ Filter By Itemsets) " ales Active Tickets 993 Build .msi packages for Windows and .mpkg packages for Mac OSX other 2.0 Future wish ales new 2011-10-28T11:55:53+02:00 2011-10-28T11:55:53+02:00 in the daily build VM ales Active Tickets 1011 Data Table should remember instance selections. other 2.0 Future wish ales new 2011-11-25T11:37:40+01:00 2011-11-25T11:37:40+01:00 ales Active Tickets 1015 Merge widget with ambiguous values canvas 2.0 Future wish ales new 2011-12-01T12:03:40+01:00 2011-12-06T12:19:22+01:00 "fails with an exception {{{ Code: return orange.ExampleTable([dataA, mergingB]) KernelException: 'orange.ExampleTable': Meta attribute 'clone name' has ambiguous values on example #5674 }}} This should be done more gracefully (error message, option to remove/duplicate conflicting columns) " ales Active Tickets 1042 The canvas widget menu popup should complete by widget tags/keywords canvas 2.0 Future wish ales new 2011-12-15T14:50:53+01:00 2011-12-15T14:50:53+01:00 not just widget names. ales Active Tickets 1054 Add a network output signal to widgets. canvas 2.0 Future wish janez new 2012-01-10T12:49:02+01:00 2012-01-17T23:41:29+01:00 "We would like to have network output signals for widgets with inherent network structure: Classification Tree, Random Forest, Regression Tree, RF Regression, Hierarchical Clustering, Interaction Graph, Association Rules and Itemsets ?" miha Active Tickets 1111 MultiTreeLearner fails on missing classes library 2.0 Future wish lanz accepted 2012-02-27T12:51:03+01:00 2012-02-27T13:33:42+01:00 "MultiTreeLearner fails if any of the classes includes missing values. Should work for this type of data as well. Example: {{{ import Orange data = Orange.data.Table(""bridges.mt1.tab"") tree = Orange.multitarget.tree.MultiTreeLearner() c = tree(data) }}} Throws ValueError: Data has missing class values." blaz Active Tickets 1124 Enable use of Projectors as preprocessors library 2.0 Future wish new 2012-03-08T20:05:48+01:00 2012-03-08T20:05:48+01:00 Linear projectors (Orange.projection.linear.Projector) could be used as preprocessors. matija Active Tickets 1143 SVM - custom kernels library 2.0 Future wish new 2012-03-19T13:56:00+01:00 2012-03-19T13:56:00+01:00 "Currently, after setting kernel_func it is immediately transformed into an instance of KernelFunc and getting the original kernel object or its attributes is impossible. {{{ import Orange svm = Orange.classification.svm.SVMLearner() svm.kernel_type = svm.Custom mykernel = lambda i1, i2: 0 mykernel.gamma = 1 svm.kernel_func = mykernel svm.kernel_func.gamma # does not work }}} It would be useful to be able to access the kernel object (to see which one it is, see its parameters, ...). " lanz Active Tickets 1173 Implement a better input/output declarations for widgets. canvas Future wish ales new 2012-04-06T15:48:35+02:00 2012-04-06T15:48:35+02:00 ales Active Tickets 1182 The Canvas main function should run some checks and display canvas Future wish ales new 2012-04-23T14:57:34+02:00 2012-04-23T14:57:34+02:00 a message (using Tkinter ??) if for instance PyQt4 is not installed or incompatible sip, no pkg_resources etc. Instead of this now when the user starts it (for instance on windows) nothing happens. ales Active Tickets 1184 error function in widgets should also accept an exception instance. canvas Future wish ales new 2012-05-04T10:49:45+02:00 2012-05-04T10:49:45+02:00 The user should also be able to retrieve the full stack trace. ales Active Tickets 1231 Optimize the Predictions widget canvas Future wish ales new 2012-09-11T15:49:33+02:00 2012-09-11T15:49:33+02:00 On large input datasets it spends to much time in table layout code. Use Qt's layout optimization flags (same size etc...) ales Active Tickets 1269 Add attribute normaliation to SOM library Future wish ales new 2013-01-21T11:50:25+01:00 2013-01-21T11:50:25+01:00 Add on option to SOMLearner to perform mean and std dev. normalization of the features. ales Active Tickets 1270 Add multi color gradient for SOM component planes library Future wish new 2013-01-21T11:54:41+01:00 2013-01-21T11:54:41+01:00 As reported [http://orange.biolab.si/forum/viewtopic.php?f=4&t=1702 here] ales Active Tickets 1271 Add automatic clustering to SOM Visualizer canvas Future wish new 2013-01-21T12:19:07+01:00 2013-01-21T12:19:07+01:00 Add an option in 'SOM Visualizer' widget to cluster the codebook vectors using either hierarchical or k-means clustering (similar to [http://www.cis.hut.fi/somtoolbox/ somtoolbox] implementation). ales Active Tickets 1290 Add support for mingw32 compiler library Future wish new 2013-02-27T20:02:31+01:00 2013-02-28T11:17:39+01:00 "- Determine the compiler at setup.py runtime (use config command??). - Fix use of #ifdef _MSC_VER where applicable - linker needs to create orange.lib file so other pyds can link to it ([http://stackoverflow.com/questions/3269766/dynamically-linking-python-extension-pyd-to-another-extension see])" ales Active Tickets 1310 Remove uses of socket.setdefaulttimout library Future wish ales new 2013-05-31T11:19:06+02:00 2013-05-31T11:19:06+02:00 On some versions of Python this can break multiprocessing module ([[http://bugs.python.org/issue6056]]). ales Active Tickets 1311 Execute a saved scheme headlessly library Future wish new 2013-06-05T20:18:32+02:00 2013-06-05T20:19:55+02:00 "It would be useful to execute a saved scheme without loading it into the canvas. For example: 1) A user creates a scheme that validates a dataset. 2) The scheme automatically runs on many datasets with a shell script, continuous integration service, etc. 3) Results are collated and analyzed. This would allow non-technical users to create reusable components that could be integrated as part of a test suite or larger workflow. A good start would be providing some basic API support, with an eye to extending the functionality to the orange-canvas app later on." echlebek Active Tickets 1312 Add HDF5 file support to Orange library Future wish new 2013-06-05T20:29:10+02:00 2013-06-07T00:44:09+02:00 "[http://www.hdfgroup.org/HDF5/ HDF5] is used by many organizations for storing large datasets. Adding HDF5 support to Orange can be accomplished through the use of the h5py library, which reads HDF5 datasets into numpy arrays. 1D and 2D datasets can easily map to Orange.data.Table objects, so supporting those datasets is a good starting point." echlebek Active Tickets 1212 PyListModel should not use deprecated __getslice__, __setslice__ ... canvas Next 2.x bug ales new 2012-07-03T14:57:49+02:00 2013-04-02T12:06:32+02:00 "(in Orange.OrangeWidgets.OWItemModels) It should instead handle slice objects in {{{__getitem__}}}, {{{__setitem___}}}. See documentation of [http://docs.python.org/2/library/functions.html#slice slice] " ales Active Tickets 1273 Canvas crashes on exit when using Discretize widget canvas Next 2.x bug new 2013-01-28T17:11:56+01:00 2013-01-28T17:11:56+01:00 "After using Discretize widget and closing the window, canvas crashes with a segfault on MacOSX with stack trace {{{ Thread 0 Crashed: Dispatch queue: com.apple.main-thread 0 libSystem.B.dylib 0x00007fff8452a0b6 __kill + 10 1 libSystem.B.dylib 0x00007fff845ca9f6 abort + 83 2 libSystem.B.dylib 0x00007fff844e2195 free + 128 3 orange.so 0x0000000101076ae7 __gnu_cxx::new_allocator > > >::deallocate(std::_Rb_tree_node > >*, unsigned long) + 29 (new_allocator.h:97) 4 orange.so 0x0000000101076b0f std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_put_node(std::_Rb_tree_node > >*) + 37 (stl_tree.h:373) 5 orange.so 0x0000000101076b5a std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_destroy_node(std::_Rb_tree_node > >*) + 72 (stl_tree.h:403) 6 orange.so 0x0000000101076bc3 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 65 (stl_tree.h:1328) 7 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 8 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 9 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 10 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 11 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 12 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 13 orange.so 0x0000000101076ba9 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::_M_erase(std::_Rb_tree_node > >*) + 39 (stl_tree.h:1326) 14 orange.so 0x0000000101076bf6 std::_Rb_tree >, std::_Select1st > >, std::less, std::allocator > > >::~_Rb_tree() + 34 (stl_tree.h:594) 15 orange.so 0x00000001010e6af9 std::map, std::less, std::allocator > > >::~map() + 21 (stl_map.h:94) 16 orange.so 0x000000010106e0d2 TContingency::~TContingency() + 118 (contingency.cpp:125) 17 orange.so 0x00000001010771a9 TContingencyClass::~TContingencyClass() + 35 (contingency.hpp:90) 18 orange.so 0x0000000101077251 TContingencyAttrClass::~TContingencyAttrClass() + 35 (contingency.hpp:134) 19 orange.so 0x000000010105d038 Orange_dealloc(TGCCounter*) + 77 (cls_orange.cpp:760) 20 orange.so 0x0000000101002597 TGCCounter::freeRef() + 59 (garbage.hpp:96) 21 orange.so 0x0000000101059182 GCPtr::operator=(GCPtr const&) + 64 (garbage.hpp:246) 22 orange.so 0x000000010138e846 TOrange::dropReferences() + 82 (root.cpp:159) 23 orange.so 0x000000010105b445 Orange_clear(TGCCounter*) + 43 (cls_orange.cpp:254) 24 org.python.python 0x00000001000ff5a8 collect + 1496 25 org.python.python 0x00000001000ffc1f gc_collect + 111 26 org.python.python 0x00000001000c04cd PyEval_EvalFrameEx + 21437 27 org.python.python 0x00000001000c24d6 PyEval_EvalCodeEx + 2134 28 org.python.python 0x00000001000c0015 PyEval_EvalFrameEx + 20229 29 org.python.python 0x00000001000c24d6 PyEval_EvalCodeEx + 2134 30 org.python.python 0x00000001000c25f6 PyEval_EvalCode + 54 31 org.python.python 0x00000001000e716e PyRun_FileExFlags + 174 32 org.python.python 0x00000001000b827c builtin_execfile + 364 33 org.python.python 0x00000001000c04cd PyEval_EvalFrameEx + 21437 34 org.python.python 0x00000001000c24d6 PyEval_EvalCodeEx + 2134 35 org.python.python 0x00000001000c25f6 PyEval_EvalCode + 54 36 org.python.python 0x00000001000e716e PyRun_FileExFlags + 174 37 org.python.python 0x00000001000b827c builtin_execfile + 364 38 org.python.python 0x00000001000c04cd PyEval_EvalFrameEx + 21437 39 org.python.python 0x00000001000c24d6 PyEval_EvalCodeEx + 2134 40 org.python.python 0x00000001000c25f6 PyEval_EvalCode + 54 41 org.python.python 0x00000001000e716e PyRun_FileExFlags + 174 42 org.python.python 0x00000001000e7429 PyRun_SimpleFileExFlags + 489 43 org.python.python 0x00000001000fe35d Py_Main + 3085 44 python 0x0000000100000f14 0x100000000 + 3860 }}} This only happens if changing the selected attribute in the 'Individual attribures' frame (using 'wine.tab' data set) " ales Active Tickets 772 orngCI other 2.0 Next 2.x task blaz assigned 2011-05-03T11:45:40+02:00 2012-02-07T21:49:55+01:00 marko Active Tickets 775 orngInteract, orngInteractions other 2.0 Next 2.x task blaz assigned 2011-05-03T11:48:32+02:00 2012-02-07T21:51:52+01:00 marko Active Tickets 778 orngPade other 2.0 Next 2.x task crt accepted 2011-05-03T11:51:00+02:00 2012-02-07T22:21:37+01:00 marko Active Tickets 783 orngMosaic other 2.0 Next 2.x task miha assigned 2011-05-03T11:52:51+02:00 2012-02-07T22:22:30+01:00 marko Active Tickets 785 orngContingency, orngCRS, orng2Array, orngDimRed other 2.0 Next 2.x task marko new 2011-05-03T11:58:40+02:00 2012-02-07T22:22:08+01:00 Should we transfer them? marko Active Tickets 1079 document Orange.statistics.evd other 2.0 Next 2.x task new 2012-02-04T11:50:57+01:00 2012-02-07T22:00:33+01:00 marko Active Tickets 1123 Refactor create_pls_projection into a class (if applicable) library 2.0 Next 2.x task new 2012-03-08T20:01:41+01:00 2012-03-08T20:01:41+01:00 Refactor Orange.projection.linear.create_pls_projection into a Pls class that returns a projector - if feasible. matija Active Tickets 1090 associate rules from sparse data in tab file addon/bioinformatics 2.0 Next 2.x wish janez new 2012-02-08T10:17:01+01:00 2012-02-08T10:17:01+01:00 "The output is strange and includes tuples ""Bread=#RNGE"". The script: {{{ import Orange data = Orange.data.Table(""market-basket.tab"") rules = Orange.associate.AssociationRulesSparseInducer(data, support=0.3) print ""%4s %4s %s"" % (""Supp"", ""Conf"", ""Rule"") for r in rules[:5]: print ""%4.1f %4.1f %s"" % (r.support, r.confidence, r) }}} outputs {{{ Supp Conf Rule 0.6 0.8 Bread=#RNGE -> Milk=#RNGE 0.6 0.8 Milk=#RNGE -> Bread=#RNGE 0.4 0.5 Bread=#RNGE -> Milk=#RNGE Diapers=#RNGE 0.4 0.7 Bread=#RNGE Milk=#RNGE -> Diapers=#RNGE 0.4 0.7 Bread=#RNGE Diapers=#RNGE -> Milk=#RNGE }}} " blaz