Back to the main page.

Bug 2237 - rollback_provenance: fails when data structure lacks a cfg-field that is a structure

Reported 2013-08-08 21:34:00 +0200
Modified 2014-01-29 13:28:39 +0100
Product: FieldTrip
Component: core
Version: unspecified
Hardware: PC
Operating System: Mac OS
Importance: P3 normal
Assigned to: Eelke Spaak
Depends on:
See also:

Jan-Mathijs Schoffelen - 2013-08-08 21:34:45 +0200

I am looking into some failing test functions on the dashboard at the moment. For some of the test functions we create some dummy data structures, that typically don't contain a cfg-field. I encountered now 2 situations where the test function fails because rollback_provenance expects 1) a cfg-field to be always present 2) the cfg-field to be a structure (or perhaps config object is allowed, I don't know for sure). Question: should we catch this in rollback_provenance? or do we expect all data structures to have a cfg-field (as a struct)?

Jan-Mathijs Schoffelen - 2013-08-08 21:38:12 +0200

PS: an example is test_bug2222, where I added freqX.cfg = struct([]);

Eelke Spaak - 2013-08-09 09:19:57 +0200

I think rollback_provenance should fail gracefully (i.e. do nothing) when there is no cfg field in the data. Making a cfg field required is pretty darn ugly.

Eelke Spaak - 2013-08-21 14:29:20 +0200

fixed in r8407

Eelke Spaak - 2014-01-29 13:28:39 +0100

changing lots of bugs from resolved to closed.