Prom-lite bugs
 Share
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

 
$
%
123
 
 
 
 
 
 
 
 
 
ABCDEFGHIJKLMNOPQRSTUVWXYZ
1
Plug-inTesterDescription of the bug
Status (Opened vs Closed vs Unsolvable for this release)
DeveloperNotes
2
Replay for ConformanceAlifahIn "Set parameters" it would be nice if the transition and event class are ordered alphabetically. The user may change the cost for a specific transition/event class, so it will be easier to find it if they are ordered.Solved (and tested by Alifah)EricPNetReplayer, version 6.5.75, revision 28456. Seems to work for the Conformance/Performance replayer as well. both must use the same dialog here.
3
Inductive visual MinerAlifahI tried Inductive visual Miner with MXML classifier and it shows some deviation. However, using the PetriNet Replayer the fitness value is 1.0. It would be nice if there is a description that explains the alignment between those two has different semantic.Solved (and tested by Massimiliano)SanderI've removed the possibility to use a lifecycle:transition classifier from IvM.
4
Replay for Conformance/PerformanceAlifahIn the default settings: if there are some invisible transitions, the next transition will have NaN as the average sojourn timeUnsolvableThis requires an understanding of the code of the replayer.
5
Replay for ConformanceAlifahMap transitions to event classes is missing if there is already the object of PNRepResult. One needs to delete the connection files before she/he can do the mapping.UnsolvableEric
6
Replay for Conformance/PerformanceAlifahThere is a strange start activity which does not mapped correctly. This is not happen in the complete activity and the other activities. Picture: https://drive.google.com/file/d/0B7XrNQsrc0HgOXVIZ0lFR2pFRjQ/view?usp=sharingUnsolvableEric: May be the result of the 'fuzzy' mapping as used by Arya.
7
CSVImporterBartImproper CSV file with missing value can lead to exception being shownClosed (and tested)FelixFixed. The CSV file does still not load as it has an inconsistent number of columns throughout the file, but NULL header fields are handled correctly and the error messages have been improved.
Potential feature request: add possibility to skip unconforming lines during import? Actually that is what I expected from the options in the dialog.
8
CSVImporterBartIt is possible to select a column multiple times when selecting case or event column. Is this wanted behavior?Closed (and tested)FelixYes. This leads to the values being concatenated.
9
CSVImporterBartSome typos (e.g. - XFactory: "XFactory implementation that is be used to..."
- attribute conversion mode: "Add attributes to either to both...")
Closed (and tested)FelixFixed. Thanks
10
Importing Multiple FilesBasWhen importing multiple files of multiple types, by dragging them into the workspace, ProM sometimes totally crashes. No error is shown in cmd, however, ProM is completely frozen. The bug only seems to occur whenever I did not import anything else. If i already imported something via the import button, it does workclosed (and tested)EricThe initial release did not contain the new UItopia
11
Testing remembering last visualizer
BasDoes not seem to work at all... I imported a log, switched the visualizer to dotted chart.
Closed all visualizers, clicked the eye button again, though I got the log dialog again.
Also, importing another log gives me the standard log dialog.
closed (and tested)EricThe initial release did not contain the new UItopia
12
TAR.GZ archive (all platforms)EduThe package is not compressed with GZip, it is just a tar.Closed (and tested)EricPerhaps if you use the "z" flag for tar (?). Or perhaps modern tar is smart enough to figure out what to do with a gzipped tar even without the "z" flag. Talked with Edu. It seems Chrome automatically unzips the gzipped file without changing the name. Apparently it is a known unresolved bug: https://bugs.chromium.org/p/chromium/issues/detail?id=47951
13
Multi-perspective ExplorerEduIn the "Performance & Alignment-related Parameters" window, the labels of the options are too long and do not fit in the window.Closed (and tested)Felixhttps://www.dropbox.com/s/q169el096na21mc/label-names-parameters.png?dl=0Fixed
14
Multi-perspective ExplorerEduWhen exploring the traces and alignments, it would be nice to have an option in the right click menu to select all the traces instead of having to select one by one to export them.Closed (and tested)Felixhttps://www.dropbox.com/s/7w4qzs4m93zi3cu/select-all.png?dl=0Standard keyboard shortcuts work. For example CTRL+A to select all. I don't think adding more features helps much.
15
Multi-perspective ExplorerEduThe labels of some activities overflow the size of the boxNot solvable with reasonable effortFelixhttps://www.dropbox.com/s/zreutwvgq35vma6/labels-out-of-boxes.png?dl=0Just happens on Linux and has something to do with Fonts. I need a Linux test system to investigate.
16
Multi-perspective ExplorerEduWhen discovering guards, some labels are missplaced (I think) like the one in the image in which two "false" guards are placed next to a single edge. Probably one of them refers to the edge below.UnsolvableFelixhttps://www.dropbox.com/s/tohssoof5p1t5wj/double-label-false-location.png?dl=0This is what I get from GraphViz. Nothing I can do about :(
17
Replay a Log on Peti Net for Conformance AnalysisEricAfter having fixed bug 16, another bug surfaced: The replayer now runs out of the stack. Closed (and tested)EricDiscussed this with Boudewijn. Some part of the replayer is recursive, and by chance with this the recursion depth required is approx. the number of traces in the log. Hence, about 10000. Boudewijn has replaced the recursion with a normal loop, but we need to test whether the results of the replayer have not chenged as a result of this replacement. Otherwise, we leave the replayer as is, which causes problem with nets like this (nice catch, Felix!). In practice, this bug is unlikely to show. Who in his right mind is going to replay 1000 traces on a net that contains no activities? Fixed by revision 6.5.66 of PNetReplayer. This is awkward. Somebody has been adding/revoming lines here. As a result, my numerical references (like 16, meaning the bug reported on line 16) to other bug fixes now are pointing to unrelated bug fixes. This shows that next time we should not use a spreadsheet like this.
18
Inductive MinerEricInductive Miner - exhaustive K-successor takes ages on the BPIC15_1 log, without showing progress.Solved (and tested)SanderSolved by asking confirmation when about to start these two variants on logs with many activities.
19
Inductive MinerEricInductive Miner - Incompleteness takes ages on the BPIC15_1 log, without showing progress.Solved (and tested)SanderSolved by asking confirmation when about to start these two variants on logs with many activities.
20
Inductive MinerEricInductive Miner - lifecycle crashes on the BPIC15_1 log.Solved (and tested)SanderArrayIndexOutOfBoundsException: -1 at org.processmining.plugins.InductiveMiner.graphs.GraphImplLinearEdge.getEdgeSource(GraphImplLinearEdge.java:170)
21
Inductive MinerEricInductive Miner - infrequent & lifecycle crashes on the BPIC15_1 log.Solved (and tested)SanderArrayIndexOutOfBoundsException: -1 at org.processmining.plugins.InductiveMiner.graphs.GraphImplLinearEdge.getEdgeSource(GraphImplLinearEdge.java:170)
22
Replay for Conformance/Performance
EricThe mapping from transition labels to activities does not work like it should. Identical labels are not matched.Solved (and tested)EricFixed in revision 28164.
23
Replay for ConformanceEricChanging costs for model moves is a pain-in-the-ass when
invisible model moves are present. Ideally, the user can set the costs
for visible model moves in one go without changing the costs for invisible
model moves, and vice versa. That is, the plug-in should distinguish visible
model moves and invisible model moves.
UnsolvableEric
24
Replay for ConformanceEricSetting costs for transitions with identical labels does not really work, as the user cannot tell for which transition s/he is setting the costs for. Better to set costs for the labels (like the Replay for Conformance/performance plug-in does).UnsolvableEric
25
Replay for ConformanceEricCosts can be set to string like “Hank” (treated as 1?). Not a big issue, but from a usability perspective this is odd.UnsolvableEric
26
Replay for Conformance/Performance
EricA lot of dialogs are shown to the user, some of which make no sense to him/her (at least, not to me). There should be a dummy mode where only the classifier, the mapping, and the costs can be set, and defaults are used for the others. Unsolvable
27
Replay for Conformance/Performance
EricThis plug-in only makes sense if the log contains timestamps. Either the log contains time:stamp as global event attribute and that is used, or the user needs to specify a global timestmap attribute. This should be started at the beginning, not somewhere half-way. If not timestamp is present, this plug-in could 'degrade' to the Replay for Conformance plug-in.Unsolvable
28
Visualize Petri net (DOT) GraphVisualizersFelixThe visualizer shows a strange name "sour... \n source 1" for the source place of this Petri net (https://www.dropbox.com/s/r3vlgeyh5agh45h/Net1.pnml?dl=0) . The correct name is "source 1". Similarly for the sink place it shows "sink ..." even though the real name is "sink 2".Closed (and tested)EricRevision 28162. The lables of places are now shown outside the place, which also prevents the place from blowing up as a result of a long place label. Tooltips are now not copied into the label (this is the "\nsource 1" part Felix is talking about).
29
Replay a Log on Peti Net for Conformance AnalysisFelixAfter importing this Petri net (https://www.dropbox.com/s/r3vlgeyh5agh45h/Net1.pnml?dl=0) and starting conformance checking to this log (https://www.dropbox.com/s/aeb3lf2s5l662s8/combined_log_15_percent_noise_max_length35.xes.gz?dl=0). Aryas replayed crashes with a NPE (https://www.dropbox.com/s/97g8j44pnu7asz7/replayer-stacktrace.txt?dl=0).Closed (and tested)EricRevision 28149. There was a problem with the annotations.
30
Framework / Load PNMLFelixAfter importing this Petri net (https://www.dropbox.com/s/05g08rmbgl811cw/resource-list-stacktrace.txt?dl=0) the UI becomes unresponsive because of a NPE in the ResourceListCellRenderer. Stacktrace (https://www.dropbox.com/s/05g08rmbgl811cw/resource-list-stacktrace.txt?dl=0)Closed (and tested)EricRevision 28149. There was a problem with the annotations.
31
Import FrameworkFelixAfter importing a Petri net the Petri net does not always appear in the list of resources of ProM. This seems to happend randomly.Cannot be reproduced (tested, and I cannot reproduce it either maybe related to another issue)EricFelix, could you check whether there is a new XLog object in the All tab? In case of problems with the PNML file, a log will be generated explaining the problems.
32
Import Petri net as DPNFelixImporting a Petri net saved by the normal PNML exporter as DPN does not import the final marking saved by the normal PNML exporter. This is because the DPN implementation uses (has used for a while) a different mechanism to store the final marking. This is not so nice :)UnsolvableFelixEric: I guess this requires a change in the DPN import? Felix: I think that this needs more testing/consideration. So I would rather not try to fix this now. I reported it more that we do not forget about it.
33
Inductive Miner (IVM)FelixException "java.lang.IllegalArgumentException: Cannot pass as a parameter to a plugin" show when clicking on 'Select visualization' -> 'Inductive Visual Miner' while the IVM is already running: https://www.dropbox.com/s/2h436t2qznc0hg3/IVM-Bug.png?dl=0Solved (and tested)SanderThe launcher object now remains in ProM, which fixes this issue. The log is referenced in this launcher object using a soft reference.
34
ProMLiteJoosWhen starting from the .bat file I cannot drag&drop files (independant of java version). It does work when starting the .exe file. No errors reported. rev. 28149. Others (Eric, Bart) do not seem to have this issue. No errors are reported...Closed (and tested)EricThis is caused by the system property suppressSwingDropSupport. If set to true, drag and drop will not work. The bat and exe file will set the property to false.
35
XESLiteJoosMany 'xesliteDDDDDD.db' files are left in the 'C:\Users\jbuijs\AppData\Local\Temp' directory (on Windows)UnsolvableFelixThese files should, normally, be removed after closing ProM. I just tried to reproduce the problem, but all files were removed. In some cases this is not possible. For example, when ProM is forcefully closed. There is a known issue with MapDB that in some cases the files cannot be deleted. Therefore, these files are created in the TEMP folder, which is cleared by the normal Windows disk clean-up tool.
36
ProMLiteJoosAfter removing ProM (in my case ProMLite 1.1 from April 29) the .ProMLite11 folder was not deleted, hence packages remainedUnsolvableEricThe uninstaller only removes the framework, as it is completely unaware of any packages that were installed by ProM.
37
CSVImporterJoosImported CSV files cannot be visualized (while the importer plug-in can do this)To be postponedFelixThe CSV Visualizer is not yet part of ProMLite. So it would be a matter of adding the annotations. I would like to leave this decision to Eric and Massimiliano.
38
CSVImporterJoosImported CSV files cannot be exported to diskTo be postponedFelixThe CSV Exporter is not yet part of ProMLite. So it would be a matter of adding the annotations. I would like to leave this decision to Eric and Massimiliano.
39
Inductive visual MinerMarcoMaking the ProM window too small (or starting it without redrawing/revalidate/repaint) gives an ArraySizeNegativeExceptionSolved (and tested)SanderA check is included to check for non-positive window widths/heights. Then the animation is paused and no histogram is shown.
40
Multi-perspective ExplorerMassimilianoI get “org.processmining.xeslite.query.syntax.ParseException: Missing query after '%'” if, in the trace view, I press enter on the filter field and the filter field is emptyclosed (and tested)FelixFixed
41
Multi-perspective ExplorerMassimilianoIf I choose “Display Process Model”, the trace view is empty. Is it intended?closed (and tested)FelixFixed
42
Data precisionMassimilianoThe precision panel is hidden. One needs to manually open it. Bug pointed out by a student who could not find the precision valueclosed (and tested)FelixFixed
43
Data precisionMassimilianoTo compute the precision, all log attributes can be considered. Shouldn't the plug-in only allow one to select the attributes that are belonging to the modelclosed (and tested)FelixNo, I don't think so. Precision is defined over all attributes of the log. Of course you can manually decide that some are not worth considering in the configuration and we omit the timestamp and such attributes. If you deselect in the configuration all 'relevant attributes', then, only the variables are taken into account. --- To prevent confusion of users none of the relevant attributes will be selected by defautl. -- Fixed
44
Data-aware Conformance CheckingMassimilianoThe visualization of the alignments show "fitness=1" even with moves in the logs. As a matter of fact, those moves in the logs refer to events mapped to no model transitions. So, it's normal that fitness is 1 but the visualization is counterintuitive. Should the log moves for those events not shown at all?closed (and tested)FelixThis is the exact result of the alignment. Unmapped events are associated a cost of 0 by default. So fitness is 1.0 if there are log move only for those events. I would not want to hide those, because their existence is useful information. Maybe we could add a toggle in the cost configuration to switch between the mode 0 cost for unmapped events and 1 cost for unmapped events. ---- Set the default cost back to 1 as the cost for those events is not taken into account anyway.
45
Data-aware Conformance CheckingMassimilianoWhen in the alignment group by variant, I specify filter "fitness < 1", I am shown no traces, even though some traces existclosed (and tested)FelixFixed
46
Data-aware Conformance CheckingMassimilianoWhen in the alignment group by variant, If I make the filter field empty and press ENTER, I expect to see no filters. closed (and tested)FelixFixed
47
Multi-perspective ExplorerMassimilianoThe comment “In the trace view, when I click on the background, all trace variants should be unselected. As a nice consequence, the main screen does not highlight any path in blue” seems not to be addressed. There is no way to deselect all traces in the trace viewclosed (and tested)FelixQuite difficult to implement, as the white space in the trace view is not the background, but belongs to the displayed item. CTRL+Click also deselects like in the Windows Explorer.
Create a button to clear the seelction (or something similar)
48
ILPMinerNourIn the configuration panel, if I choose either dual or hybrid with the same configuration for other parameters, I expect having the same result. The difference is only in the performance. Is that true or I am missing something? I got different results when taking a log with two traces: saab and saba and these configurations: https://drive.google.com/folderview?id=0B9BtFuP8jjVnOVZoRVBBTmthUk0&usp=sharingNothing to be doneBasI'm curious what the different results are. Note that theoretically there might be several optimal solutions having the same, optimal, score. Using hybrid versus dual, in this case implies a different number of variables within the underlying ILPs. This in turn can lead the solver to find different results which are both correct.
49
ILPMinerNourAlso related to sequence encoding filter, setting the cut-off to 0 means that only edges with the maximal value are kept and all others are removed. Taking a log with abcd (100 times) and acbd (1 time) and a cut-off 0 should lead to abcd, however I am getting a different result. I am following the algorithm as explained in the article. Maybe something missing regarding the cut-off meaning? (the log and configurations can be found here "https://drive.google.com/open?id=0B9BtFuP8jjVnYmVxM0czdjBGWjQ")Unsolvable for this releaseBasProbably this is due to the fact that the causal graph used is not aware of the filtering. Hence the causal graph can find causalities that would no longer be within the filtered constraint body. I reused some code of Eric here, both because he already implemented a lot of causal graphs and I implmeneted it before I came up with the filtering. I didn't feel like reimplementing the causal graphs based on the sequence encoding graphs. Technically I don't really consider this as a bug, though I agree that it is a very inconvenient "feature".
50
ILPMinerNourI didn't found how the filter "Slack variable filter" worksNothing to be doneBasThere is a technical report on bpmcenter.org detailing on this kind of filtering.
51
ILPMinerNour(minor, usability) It is better to add an explanation in the configuration panel that a right click opens a previously used configurationClosed (and tested)Bas
52
ILPMinerNourIt is unclear how the sequence encoding filter works. For example when I set the cut-off to 1, I expect having the same result without filtering. This is because the difference (as explained in the article) will be always 0 and therefore there are not cuts.Nothing to be doneBasThis should indeed be the case. Do you have any log where this does not seem to be the case.
53
ILPMinerNour(minor, usability) It would be nice if there are some mouseover explanations for the different types of causal relations. Just a general indication for a one who does not need to understand them in detail.Unsolvable for this releaseBasI agree, however, I should discuss this with Eric as I am reusing his code here.
54
Framework / UItopiaSanderMaking the ProM window too low (i.e. drag the bottom border all the way up) throws an exceptionSolvedEricSolved by revision 28414.
55
Inductive MinerEricImport the BornToFail_Event_MissingAllNames.xes file. Start the Inductive miner with default parameters (including Activity classifier, which is defined in the log). The classifier attribute seems not to be used (is concept:name being used in the background?), as it the only transitions contains an empty name. The InductiveVisualMiner does this better.Solved (and tested)SanderHV: THis is caused by the fact that the "Activity" classifier seems to be preselected, but is not. As a result, I guess the default classifier "concept:name" is used instead, as nothing has been selected...
56
Inductive visual MinerFelixToken cannot move back in time exception for some logs (e.g., https://www.dropbox.com/s/aeb3lf2s5l662s8/combined_log_15_percent_noise_max_length35.xes.gz?dl=0) causing this stacktrace: https://www.dropbox.com/s/kksnprkvezlv4mj/ivm-stacktrace.txt?dl=0Closed (and tested)Sander
57
Multi-perspective ExplorerEduNullPointerException when using default parameters on "(Re)compute Alignment" with log exercise6.xes.zip from sample logs. Using a different classifier solves the problem. (Details emailed to Felix already)ClosedFelixFixed. That was a 'evil bug'. Log move costs were calculated using a different classifier. Thanks for finding :)
58
Conformance Checking of DPNAlfredo· In the Projection to net visualization, if you right-click on e.g., the fist transition of the model, and then select the option: Show alignments containing activity X, it shows no alignments.ClosedFelixFixed all of these issues. Thanks for pointing out.
59
Conformance Checking of DPNAlfredo· In the Projection to net visualization, if you right-click in a transition and then select the option “Show alignments with violations for activity x” in the popup visualization , if you click on the “approximate matches” button (bottom of the screen) twice (i.e., select and deselect) you get an ugly popup with: org.processmining.xeslite.query.syntax.ParseException: Missing query after '%'ClosedFelix
60
Conformance Checking of DPNAlfredo· In the Projection to net visualization, if you double click on a node, the popup window showing a tree has an html code text.ClosedFelix
61
Conformance Checking of DPNAlfredoIf you visualize alignments and click on “sub-string matches” alignments change, but when you click again on it (re-enable) the alignments do notClosedFelix
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
Loading...