Back to the main page.
Bug 1804 - check whether ft_defaults respects existing path-order
| Status | CLOSED FIXED |
| Reported | 2012-10-31 13:18:00 +0100 |
| Modified | 2013-06-05 12:15:40 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Windows |
| Importance: | P3 normal |
| Assigned to: | Roemer van der Meij |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Jan-Mathijs Schoffelen - 2012-10-31 13:18:22 +0100
Roemer knows what to do...
Roemer van der Meij - 2012-11-08 12:27:24 +0100
Fixed for now, but let's discuss implementation. For several sub-folder/modules, there was a call to ft_hastoolbox in ft_defaults, but there was no actual check being performed by ft_hastoolbox. This resulted to several sub-dirs always being added to the path, regardless of whether they were already on it. Hence, they're always at the top of the path. The sub-dirs/modules in question: - trunk/preproc - trunk/inverse - trunk/realtime - trunk/specest - trunk/statfun - trunk/trialfun The first 4 are actual modules (as documented on the wiki), the last two are more a kind of 'utilities' The current structure of ft_hastoolbox distinguishes toolboxes from 'other (sub)directories in fieldtrip'. Modules are considered toolboxes. For a toolbox, an existence-on-the-path check is performed by testing the existence of several files on the path, usually 2 or more ft_ functions. For a 'other (sub)directory in fieldtrip', an existence-on-the-path check is performed by checking whether the dir in question is on the current matlab path (by parsing the output of 'path'). For now, I have added the above modules/dir to the path in way identical to the last situation. Even though some of them are modules. To discuss: For modules, what is the rationale for checking the existence of an arbitrary set of main-functions instead of the sub-directory. Checking for sub-directories seems safer, as it is more stable. PS: A part of the cause for all fieldtrip-dirs being put on the top path everytime ft_defaults was called, is that their existence was only checked for the trunk folder being named 'fieldtrip'. I changed this so that it used the actual trunk (like fieldtrip-XXX; sensitive to \ and / of course) Also, there was a typo in the case for FORWARD(was FORMWARD). Also, I updated some of the urls to all modules supported in ft_hastoolbox. (multivariate and qsub are not supported in hastoolbox, and spike is missing on the wiki)