Back to the main page.
Bug 2655 - rollback_provenance interferes with configuration tracking
| Status |
NEW |
| Reported |
2014-07-17 09:17:00 +0200 |
| Modified |
2015-02-10 15:20:42 +0100 |
| Product: |
FieldTrip |
| Component: |
core |
| Version: |
unspecified |
| Hardware: |
PC |
| Operating System: |
Mac OS |
| Importance: |
P5 normal |
| Assigned to: |
|
| URL: |
|
| Tags: |
|
| Depends on: |
|
| Blocks: |
|
| See also: |
http://bugzilla.fcdonders.nl/show_bug.cgi?id=2054http://bugzilla.fcdonders.nl/show_bug.cgi?id=2832 |
Robert Oostenveld - 2014-07-17 09:17:58 +0200
when running a test script yesterday with the (for me default) cfg.trackconfig=on I noticed an error message about "cfg.trkcfgcount" in ft_checkdata which was caught in ft_preamble and hence not escalating.
The specific problem is in configuration tracking, but there might also be something affecting other information that we temporarily store in the cfg, such as the warning_once states.
A specific instance of the problem can be reproduced by adding
if ~isfield(cfg, 'trkcfgcount'), keyboard; end
twice to utilities/ft_checkconfig around line 555.
What happens there is that the trkcfgcount field is added to keep track of something (just like some other fields are added elsewhere, e.g. for the warnings). I suspect that those fields get removed/updated/messed-up by the call of an initial high-level FT function to another high-level FT function, followed by the call to rollback_provenance. The consequence is that the initial high-level FT function then looses that info.