Activity for gbm
-
gbm posted a comment on ticket #427
Attached patch fixes the problem.
-
One way to avoid the splitting of datetime values into two columns is to use ISO-8601 format for copy/paste. See the attached files for an example.
-
I have no problem in reading your 'Exercise2.sciprj' file with the linux version of SciDAVis (2.4.0 in Fedora). But I also couldn't open the file with the windows version of SciDAVis. Can you please test if you can open the attached file 'Exercise2-corrected.sciprj'? If it works, then to get around this bug you should avoid saving a project with minimized Table, Graph, or Matrix windows.
-
You should not use 'e' or 'pi' as a parameter name as they are defined as constants. See SciDAVis manual Miquel
-
See liborigin-devel mail list. As for the error when importing a .opj file, you can try if opj2dat from liborigin can read that file. If not, post a message to liborigin-devel attaching the problematic file.
-
Since release 1.23.1, copy/pasted data in a column is treated as numeric IF and ONLY IF all the pasted data is numeric. Can you please share a version 1.21 "*.sciprj" file BEFORE pasting the excel data and the corresponding version 2.3.0 "*.sciprj" obtained AFTER pasting?
-
I cannot reproduce this bug in linux. I guess it is a Windows only problem. As a work around you can use version 1.26
-
You can get libqwt5-qt5 .deb file from https://software.opensuse.org//download.html?project=home%3Ahpcoder1&package=libqwt5-qt5 or compile it yourself from https://github.com/gbm19/qwt5-qt5
-
Attached patch fixes it. The bug only triggers when a graph plots data from a Table Column that has the underscore "_" character in its name. Until the patch makes it to a release you can get around it by avoiding underscores in Table Column names.
-
I can't reproduce the bug in Fedora30 with neither version 2.3.0 nor version 2.1.4 of SciDAVis. Could you please describe your steps with more detail?: Create a table with a curve [you probably mean two columns 1(X) and 2(Y), from "Import ASCII..."?] Plot the curve [which plot type? Line, Scatter, Bars ... ] Make the fit [which fit type? Linear, Polynomial, Fit Wizard (which expression?) ...] Save the file. [Save it as "bug402-step4.sciprj"] Close it Reopen it. Import again the curve. Save [Save...
-
@Koushik, here is a work around for your issue a): After loading the opj file, use the 'Project Explorer' (Ctl+E) to identify all the windows with a multiline 'Label'. For the file '0-it_crashes.opj' in https://sourceforge.net/p/scidavis/scidavis-bugs/403/ these are 'Graph1', 'Graph2', and 'Graph3'. All three have a 'Label' with an empty second line. Use "Menu/Windows/Rename Window" to edit those 'Label's and make them single-line. Save the project in a sciprj or compressed sciprj file. That file...
-
I cannot reproduce the bug in SciDAVis 2.3.0 (Fedora 30). When opened in SciDAVis the uploaded file shows two graph windows with missing curves, but they are already missing in the file. My guess is that Table(s) with the curve data were deleted before saving the file. If that is not the case, could you please describe an action sequence that starting from a new empty project produces a bad '.sciprj' file?
-
Attached patch fixes it.
-
This is not a bug for me. It is more the expected behaviour given the y-values data range.
-
This is a well known feature/glitch, see bug #382. To allow for improved color handling, the format used to store SciDAVis settings was changed between versions 1.24 and 1.25. If you have problems with the workspace or table colors, you can change them with Edit/Preferences/General/Colors, Edit/Preferences/Tables. Alternatively, you can reset SciDAVis to default settings: In linux remove the file ~/.config/SciDAVis/SciDAVis.conf In windows use regedit to remove SciDAVis related settings from the...
-
Attached patch fixes the bug
-
Could you please submit a proper bug report via https://sourceforge.net/p/scidavis/scidavis-bugs/new/ ? Thanks, Miquel
-
Attached patch, applied to classdesc-python branch, commit b6e3012f, may solve the problem.
-
Attached patch solves the problem
-
Attached patch solves the problem.
-
This problem appears when you save the project while a graph window is maximized or minimized. To avoid it, you have to set all graph windows in normal state before saving the project. Alternatively you can uncheck the "Scale Fonts" checkbox in the "Edit/Preferences/2D Plots/Options" page.
-
Hi Richard, Fellype, An alternative method to normalize a column is to use the "Column Statistics" Table. See attached example.
-
This is already possible. See attachments. Note that all colors in SciDAVis have transparency level (alpha channel).
-
For "openSUSE Tumbleweed" (20190923) I was able to use python with OBS packages. Either with sudo zypper install python2-qt4 scidavis-1.25-7.1 or with sudo zypper install python2-qt5 scidavis-1.25-9.7
-
I am able to use the OBS SciDAVis 1.D25 packages with python in "Ubuntu 19.04" and "Ubuntu 18.04" (64 bits) with: sudo apt-get install python-qt4 Maybe a similar command solves the problem in OpenSuse
-
Attached patch fixes the bug.
-
Please check if attached patch fixes the bug also for Qt4-python2
-
Fellype, The patch works for me in a Qt5.12.4-python-3.7.4 build It works partially in a Qt4.8.7-python-2.7.16 build. Partially means it works for importing into new Table, not for Overwrite current table. I assume that you use Qt4-python2. Can you check if the patch works for importing into new Tables?
-
Try attached patch
-
Try attached patch
-
It doesn't crash for me either, but it was not working as intended. Please try attached patch.
-
Fellype, Can you please check if attached patch solves your problem.
-
patch -p1 < fix-sf380-does-not-load-colors-for-text.patch should work. That removes the 'a/' and 'b/' from file paths
-
Attached patch solves the problem
-
Attached patch solves the problem
-
SciDAVis developers use git (https://git-scm.com/doc) to manage code source in the github repo (https://github.com/highperformancecoder/scidavis) Info on how to use patches with git is available at https://git-scm.com/book/en/v2/Appendix-C%3A-Git-Commands-Email and https://git-scm.com/book/en/v2/Distributed-Git-Maintaining-a-Project#_patches_from_email
-
Attached patch solves the problem
-
The patch has been merged in recently released SciDAVis (version 1.25). Please, check if version 1.25 fixes the bug.
-
Attached patch fixes the problem
-
Meanwhile, there is a slightly better workaround than that mentioned by David. There is no crash when adding new curves and modifying its "Plot Associations" provided you remove all 'fit (function) curves' from plot before changing plot associations.
-
I can reproduce the crash (Fedora 28, python 3.6.8, Qt 5.11.3) The attached patch fixes the problem.
-
Attached patch fixes the problem.
-
Try attached patch
-
That seems to be a Qt bug: https://bugreports.qt.io/browse/QTBUG-70306 Attached patch recovers previous SciDAVis behaviour.
-
Hi Fellype, Can you check if the attached patch (against current HEAD ffc276e) fixes this bug.
-
There is a bug in MultiLayer::arrangeLayers (graph.arrangeLayers in python). The method fails to produce reasonable "Automatic layout" arrangements for more than 3 layers. The bug is hit both from python scripts and from interactive GUI. To reproduce in the GUI: 1. Open a new project (Ctrl+N) 2. Create a new graph (Ctrl+G) 3. Add 3 layers to the graph (3x Alt+L), accepting the "Guess" button when asked for the origin of the new layer. 4. The resulting multilayer graph is empty, as is the exported...
-
No. It was fixed in 35b4c6e
-
The attached patch fixes the bug in Qt4.
-
Yes, you can close it. It was fixed by 219cb6b.
-
Remove exit() calls
-
This is a Qt4 only bug, probably due (not confirmed) to a bug in Qt4 version of QMdiArea. Qt5 version of scidavis does not have this bug. I strongly encourage you to compile/use the Qt5 version. It should have less bugs and, in my system, it runs the regression test more than 4 times faster than the Qt4 version.
-
You can use the attached patch. Don't forget to add the generated liborigin.pc to the liborigin-devel package.
-
Line color is preserved, but gets overwritten by a rectangular frame, as you can check by exporting to SVG and editing the obtained file. Attached patch solves the problem.
-
You can use: g++ -o opj2dat $(pkg-config --cflags --libs liborigin) opj2dat.cpp Miquel
-
Thanks for the report. There are two remainig 'exit' calls in OriginFile.cpp that in normal use of the library should never show up. The attached patch removes them by using a better approach to handle the "No such file.." and "Permission denied" errors. Miquel
-
This bug was fixed after 1.23 release. Use beta version 1.23.1
-
Yes, there is a bug elsewhere. The graph window generates/receives a resize event when the constructor has already ended. At that point the autoScaleFonts boolean, disabled in constructor, is active again. The patch just reverts in the constructor the font resize that comes after the constructor. Until the real bug is found, the patch makes a reasonable workaround.
-
Attached patch solves the problem
-
To avoid this bug you can uncheck the "Scale Fonts" button in the Edit/Preferences/2D Plots/Options panel before creating the 2D plot windows. Alternatively, apply the attached patch that circumvents the bug.
-
There is a new scidavis-beta_1.D23.4_amd64.deb that fixes this bug and those of 1.23.2
-
This bug is the same as github issue #79 and was fixed AFTER release of V 1.23. It is solved in scidavis-beta 1.23.2 that you can download from scidavis_beta_1.D23.2_amd64.deb Note: scidavis-beta 1.23.2 has other bugs (see git repo for details).
-
Hi Fellype, Can you try attached patch and see if it fixes the bug?
-
Attached patch may fix this bug
-
Use c++11 standard to simplify logging
-
I am using version 1.23.2 Custom color for symbol edge/fill was introduced shortly after release 1.23 [1] I recommend you version 1.23.1 (beta). [1] https://github.com/highperformancecoder/scidavis/commit/78a1f7d ("Replace ColorBox by ColorButton for symbol edge and fill colors", 2018-06-07)
-
Remove config.h from devel headers
-
See attached images.
-
docs white space cleanup
-
Consistently use the override keyword. [labplot f056b6f9]
-
Attached patch should fix the icon problem. In Fedora I have no problem to change theme/style through Edit/Preferences/Style menu.
-
Attached patch fixes the bug
-
Attached patch may solve the problem
-
See fix for bug #345
-
Attached patch works for me in Fedora 28 for the four (Qt4/Qt5) + (py2/py3) combinations.
-
Confirmed. Workaround until fixed: Do left-click twice (but not doubleclick) on a window to make the explorer window get the focus. Then, 'delete' key and other shortcuts work as usual.
-
Current SciDAVis version imports correctly ASCII files that do NOT END with a newline (ASCII LF) character. The attached patch allows SciDAVis to properly import ASCII files both with or without a LF at end of file.
-
An explicit workaround for users would be to add columns of appropriate type before pasting the data. The attached patch changes column type to 'Text' if the pasted data contains non-numeric values and column type is 'Numeric'. You can use the attached 'paste-test.ods' to test the patch. A PR will follow.
-
This is no longer a bug, Try attached script. It runs without problem with current version.
-
Attached patch solves the bug for me. You can use the python script in the attached sciprj file as regression test for the bug.
-
I am now able to reproduce the bug. Until we get a fix, there are two ways to avoid the steady growth of font size when cycling the Multilayer (Graph) window through the Minimized, Normal, Maximized states: 1. Go to menu Edit/Preferences/2D Plots/Options page and uncheck the "Scale Fonts" checkbox. This is the best option for a data analysis workflow. 2. Even with the "Scale Fonts" enabled you can avoid the font size growth problem if you never switch directly from maximized to minimized (or vice...
-
Yes, I also had my doubts. I will try to explain why I choose the proposed solution. TableView and MatrixView are classes derived from MyWidget, the class used to realise children windows of d_workspace. future::Table and future::Matrix are classes derived form AbstractPart. The inheritance tree of d_view (a TableView widget) is: TableView / MyWidget / QWidget That of d_future_table (a future::Table QObject) is: future::Table / AbstractPart / AbstractAspect / QObject Thus, d_view and d_future_table...
-
I can't reproduce the problem, even when using binary from scidavis-1.23-2.fc27.x86_64.rpm.
-
The crash happens because both C++ and python try to release the memory allocated for the closed Table (or Matrix). That it occurs only after closing 10 Tables is because 10 is the default Undo/Redo history limit (Edit/Preferences/General). Thus, the first 10 Tables are not really closed but kept in the Undo/Redo stack. Attached patch fixes the bug.
-
Attached patch provides a working solution for me. Best control of data range is obtained by using the 'Step' parameter in the 'Scale' page of the 'General Plot Options' dialog.
-
It works for me (version 1.22.2) with: m=newMatrix("m1") m.setFormula("x*y+i-j") m.recalculate() Matrix values are recalculated (both in script or via "Apply" button) only at "selected" cells. Currently, there is no python API to select/deselect cells, but you can select all cells in matrix by clicking on the top-left corner of matrix area. Once selected, m.recalculate() will work.
-
Try attached patch
-
Try attached patch
-
Yes. In ApplicationWindow.cpp:303-311 initMainMenu is called before readSettings and insertTranslatedStrings. Try attached patch.
-
Try attached patch. The attached '.sciprj' file can be used for extensive test of barplot export as svg file.
-
I can partially confirm the problem in Fedora 27. Steps to reproduce: 1 Open the file https://sourceforge.net/p/scidavis/scidavis-bugs/274/attachment/error-bar-color-bug.sciprj 2 Double click inside Graph1 to get the 'Plot details' dialog 3 On the treewidget at left choose the 'yErr' curve 4 Click on the 'Color' button 5 Choose any color that does not match one of those available at the Color ComboBoxes, for instance, a color with at least one of the RGB values different from 0, 128, or 255 6 Press...
-
Hi Alexander, Fellype, Please, try attached patch. It should significantly reduce the time to Paste a large volume of data. Miquel
-
Fix parsing of annotation groups at files from early versions
-
Fix parsing note window properties at files from early versions
-
Fix parsing of window creation and modification dates
-
Fix parsing of formatAxis values
-
Change type of curpos and d_file_size to fix MSVC compiler warnings.
-
The code snapshots downloaded from sourceforge git repos do contain following info: 1. Project name 2. Git repo name 3. Git hash 4. Commit date The snapshot is downloaded as a zip file with name 'ProjectName-GitRepoName-GitHash.zip'. When unzipped, source files are extracted to the folder 'ProjectName-GitRepoName-GitHash' All the files in the snapshot have the commit date as modification timestamp. The attached patch, applied before 'prepare', allows to compile gnuplot from a downloaded snapshot,...
-
The SciDAVis code for vertical bar plots gets the bar width for a plot curve by calculating the minimum of the x-coordinate difference between consecutive curve points (i.e. consecutive rows in table). For a curve with only one row (point), there is no way to obtain that minimum, and the abs(x-coordinate) of the single point is taken as bar width. Unless a global bar width is implemented I don't see a better way to obtain the bar width for a one point curve.
-
I don't get a crash, either (Fedora, current git code). The ASSERT failure you get points to an 'out of range' error. Checking the code for bar plots I found some possible 'out of range' bugs. Try attached patch.
-
Improve reading of project files with matrices and multisheet workbooks
1 >