Back to the main page.
Bug 728 - change to ft_channelselection in SVN commit 3548 breaks some wildcard functionality
| Status | CLOSED FIXED |
| Reported | 2011-06-02 22:21:00 +0200 |
| Modified | 2011-06-21 15:52:21 +0200 |
| Product: | FieldTrip |
| Component: | core |
| Version: | unspecified |
| Hardware: | PC |
| Operating System: | Mac OS |
| Importance: | P1 normal |
| Assigned to: | Jörn M. Horschig |
| URL: | |
| Tags: | |
| Depends on: | |
| Blocks: | |
| See also: |
Eliezer Kanal - 2011-06-02 22:21:37 +0200
The code change in 3548 breaks wildcard selectors where the wildcard is in the middle, e.g. "M*1". The negation operator is being interpreted incorrectly here. See line 161, the code should be: Index: ft_channelselection.m =================================================================== --- ft_channelselection.m (revision 3613) +++ ft_channelselection.m (working copy) @@ -158,7 +158,7 @@ labelreg = labelreg | ~cellfun(@isempty, regexp(datachannel, ['^' channel{i}(1:end-1) '.*'], 'once')); findreg = [findreg i]; end - if length(channel{i})>1 && ~channel{i}(1)=='*' && ~channel{i}(end)=='*' && any(channel{i}=='*') + if length(channel{i})>1 && ~(channel{i}(1)=='*') && ~(channel{i}(end)=='*') && any(channel{i}=='*') % the wildcard is in the middle sel = strfind(channel{i}, '*'); str1 = channel{i}(1:(sel-1)); with parenthesis around the "channel{i}(1)==1" statement. Ideally, the equality comparisons should all be replaced with `strcmp` calls, which are more robust for this sort of comparison.
Jan-Mathijs Schoffelen - 2011-06-03 09:44:24 +0200
Hi Jörn, Could you look into this and fix it as suggested? Thanks, JM