|
Size: 2599
Comment:
|
Size: 3324
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 1: | Line 1: |
| == Known bugs == ---- <<BR>> === Main window and database === |
= Known bugs = == Main window and database == |
| Line 14: | Line 11: |
| * Load simultaneously datasets with slightly different sample rates: does not work | |
| Line 16: | Line 12: |
| * Popup menus and "Open file" dialog take a few seconds to appear on some XP systems. Unknown cause... * Popup menus and "Open file" dialogmay generate the following error: <<BR>>The class "!BstFileSelector" is undefined. Perhaps Java is not running.<<BR>>Workaround: Restart Matlab. * Navigator: may not work if more than one condition/subject opened at the same time, and next file do not have the same time definition.<<BR>>Only things that this prevents you to do is to review at the same time different raw files that were imported in small time blocks... So not a major issue... |
|
| Line 17: | Line 16: |
| === Figures issues (recordings / results) === | == Figures issues (recordings / results) == |
| Line 19: | Line 18: |
| * Figure menu "Set colormap max": does not work from the 3D figures with some versions of Matlab (eg. R2007a) * Cloned figures do not behave well <<BR>>Example: Clone + scouts + moving in time : currents on cortex are not updated * Display topography + display channels from the tree : crashes if there are bad channels * Channel selection on 2D Residuals : don't work properly for magnetometers only (indices issue) |
* Colormap menus do not work from the 3D figures with some versions of Matlab (eg. R2007a)<<BR>>Workaround: Use the Colormaps menu in the main Brainstorm window. * 2D/3D topography of modelized data (Fsynth) are always display with EEG colormap, even it is MEG * Modelized data / Residuals: no sensors selection available * "Save image" may capture a blank image on Linux * "Save file" do not check if file already exists on Linux |
| Line 24: | Line 24: |
| === Computation issues === | == Computation issues == |
| Line 27: | Line 27: |
| * OS_MEG crashes Matlab when scalp in not very well aligned with sensors<<BR>>Workaround: align your them well... | * Tree menu "Project on default anatomy" : cause Out of memory errors when processing FULL results (=> proceed by time blocks) * OS_MEG crashes Matlab when scalp in not very well aligned with sensors<<BR>>Workaround: register them well... |
| Line 29: | Line 30: |
| * "Project on default anatomy" cause Out of memory errors when processing FULL results (=> proceed by time blocks) | * Permutation tests do not work on "kernel only" results |
| Line 31: | Line 32: |
| === Scientific issues === | == Scientific issues == |
| Line 33: | Line 34: |
| * Rounding errors with time values: sometimes there are +/- 1 samples selected than expected | * New t-tests and old t-test do not give the same results, because the processing of bad channels is different: * Old tests: reject a channel when it is bad at least for one file * New tests: reject a channel only when it is bad in all the files |
Known bugs
Main window and database
Multiple screens: does not work in all the configurations
Cause: Too many possible configuration possible (each version of Linux and Matlab handle this in a different way)
Workaround: work only with your main screenTree is not updated well and some nodes are missing
Cause: Unpredictible order of execution of Java/Matlab operations
Workaround: Press F5 to refreshColormap sliders for contrast / brightness do not work with versions of Matlab < 7.4
Workaround: Manipulate the colormap by clicking on the colorbar and moving up/down (brightness) and left/right (contrast)- Some user preferences are not saved in compiled versions
Tree menu: "File > XTerm" does not work with all the linux distributions
- Tree menu: "Group conditions" does not update all the filename links (in results files, linkresults files, etc.)
- Display time series t=[0,end] + Display time series t=[begin,end]: time bounds are not good for the second figure
- Save figures as AVI movies: only available on Windows operating system
- Popup menus and "Open file" dialog take a few seconds to appear on some XP systems. Unknown cause...
Popup menus and "Open file" dialogmay generate the following error:
The class "BstFileSelector" is undefined. Perhaps Java is not running.
Workaround: Restart Matlab.Navigator: may not work if more than one condition/subject opened at the same time, and next file do not have the same time definition.
Only things that this prevents you to do is to review at the same time different raw files that were imported in small time blocks... So not a major issue...
Figures issues (recordings / results)
- Stat: t-values cannot be displayed on 3D MRI slices
Colormap menus do not work from the 3D figures with some versions of Matlab (eg. R2007a)
Workaround: Use the Colormaps menu in the main Brainstorm window.- 2D/3D topography of modelized data (Fsynth) are always display with EEG colormap, even it is MEG
- Modelized data / Residuals: no sensors selection available
- "Save image" may capture a blank image on Linux
- "Save file" do not check if file already exists on Linux
Computation issues
- Tree menu "Project on default anatomy" : does not work for kernel / absolute values
- Tree menu "Project on default anatomy" : does not work for kernels if there are some bad channels
Tree menu "Project on default anatomy" : cause Out of memory errors when processing FULL results (=> proceed by time blocks)
OS_MEG crashes Matlab when scalp in not very well aligned with sensors
Workaround: register them well...OS_MEG (overlapping spheres) fails when the scalp surface is not perfect (some garbage inside the head)
Workaround: do your scalp extraction with FreeSurfer- Permutation tests do not work on "kernel only" results
Scientific issues
- Grand average computation does not weight the different subjects with the number of trials for each subject
- New t-tests and old t-test do not give the same results, because the processing of bad channels is different:
- Old tests: reject a channel when it is bad at least for one file
- New tests: reject a channel only when it is bad in all the files