Back to the main page.
Bug 1171 - Update documentation of mexfiles, and check for corresponding .m scripts (autocompilation, error msg and help).
Status | NEW |
Reported | 2011-11-18 21:17:00 +0100 |
Modified | 2013-01-24 12:29:05 +0100 |
Product: | FieldTrip |
Component: | core |
Version: | unspecified |
Hardware: | PC |
Operating System: | Mac OS |
Importance: | P4 minor |
Assigned to: | |
URL: | |
Tags: | |
Depends on: | 1234 |
Blocks: | |
See also: |
Robert Oostenveld - 2011-11-18 21:17:22 +0100
Dimitri Papadopoulos wrote on Mon Nov 14 on the email list the following: ----------- Hi, File fileio/README reads: The following files were downloaded from http://www.koders.com base64.c base64.h I cannot find these files anywhere. Have they been removed? Has the source code been integrated elsewhere? For what it's worth I've found Matlab files that could have replaced that C code: external/gifti/@gifti/private/base64decode.m external/gifti/@gifti/private/base64encode.m Please fix the README. ----------- I suspect that the specific c and h files were moved to fieldtrip/src, where currently the source code for most mex files is kept and that the general ft_compile_mex compiles. It would be good to check all READMEs _and_ the wiki documentation w.r.t. the mex files, because that has changed a few times over the last few years and perhaps other sections of the docs are also inconsistent.
Boris Reuderink - 2011-11-21 11:36:32 +0100
*** Bug 1157 has been marked as a duplicate of this bug. ***
Dimitri Papadopoulos - 2011-11-21 22:33:48 +0100
I think the base64.c / base64.h files have been removed from Fieldtrip a long time ago - before the Subversion history. I grep'ed through the C files in Fieldtrip and was unable to find any occurrence of "base64". I think the original C source code had been replaced by pure Matlab code a long time ago: external/gifti/@gifti/private/base64decode.m external/gifti/@gifti/private/base64encode.m the above pure Matlab code has obviously been taken from Guillaume Flandin's site: http://www.artefact.tk/software/matlab/gifti/ which in turn uses code from: http://home.online.no/~pjacklam/matlab/software/util/datautil/
Boris Reuderink - 2012-01-17 12:18:08 +0100
Added time estimate.
Boris Reuderink - 2012-01-17 12:18:45 +0100
Updated time estimate.
Boris Reuderink - 2012-02-29 14:01:28 +0100
Removed reference to base64.{c,h}. $ svn ci Sending fileio/README Transmitting file data . Committed revision 5329.
Boris Reuderink - 2012-03-28 15:12:46 +0200
*** Bug 797 has been marked as a duplicate of this bug. ***
Boris Reuderink - 2012-03-28 15:24:10 +0200
There has been some recent work on mex files (see bug #1258). In this light, I asked at the weekly FT meeting what still needs to be done for documenting / cleaning the mex files: - Some mex files have a .m file with auto compile instructions. These should be removed, and replaced with scripts that present a meaningful error ("... not supported on platform ..."). - All mex files should have a corresponding .m files with help and above error message. - Check wiki and READMEs as stated by Robert. - The guidelines for win32/win64 are clear. Similar guidelines need to be created for Linux and OSX.
Robert Oostenveld - 2012-03-28 15:27:51 +0200
(In reply to comment #7) Regarding wiki documentation about mex files needs updating. See http://fieldtrip.fcdonders.nl/?do=search&id=mex. The latest insights are at http://fieldtrip.fcdonders.nl/code_guidelines#compiling_mex_files I don't see a reason for maintaining the page http://fieldtrip.fcdonders.nl/development/mexfiles. so that probably can be deleted the faqs should be cleaned up http://fieldtrip.fcdonders.nl/faq/compile?s[]=mex http://fieldtrip.fcdonders.nl/faq/how_can_i_compile_the_mex_files_on_64_bit_windows http://fieldtrip.fcdonders.nl/faq/matlab_complains_about_a_missing_or_invalid_mex_file_what_should_i_do Perhaps the faqs should remain as questions, but can all point to a (partial) common answer
Boris Reuderink - 2012-03-28 15:36:55 +0200
Update: auto compilation is cleanly implemented in /src/read24bit.m . It would be nice to use this form of auto compilation as much as possible (but not when it is to too difficult, e.g. with the buffer). When compilation fails, and error message with the platform should be printed for future bug reports.
Boris Reuderink - 2012-11-02 13:29:54 +0100
I am no longer working on FieldTrip. Hence, I donate all my bugs to the joint development user.