Opened 11 years ago
Last modified 3 years ago
#338 new task
Automatic comparison against previous version of ROPP
Reported by: | Ian Culverwell | Owned by: | idculv,cburrows |
---|---|---|---|
Priority: | normal | Milestone: | 12.0 |
Component: | ROPP (all) | Version: | 7.0 |
Keywords: | Cc: |
Description
Action 4 DRI ROPP 7
It is advised to systematically include automatic testing against previous version of the ROPP package for new versions to be released. Project Team to further explain their approach.
Change history (5)
comment:1 by , 10 years ago
Milestone: | 8.0 → 9.0 |
---|---|
Type: | enhancement → task |
comment:2 by , 8 years ago
Regression tests (vs ROPP8.0) passed OK at ROPP9.0.
Rolling ticket forward to ROPP10.0.
comment:3 by , 8 years ago
Milestone: | 9.0 → 10.0 |
---|
comment:4 by , 4 years ago
Milestone: | 10.0 → 11.0 |
---|
Regression tests (vs ROPP9.1) passed OK at ROPP10.0.
I really don't think there's any more we can do with this. Rolling over to ROPP-11, nevertheless.
comment:5 by , 3 years ago
Milestone: | 11.0 → 12.0 |
---|
Regression tests (vs ROPP-10.0) passed OK at ROPP-11.0.
Rolling over to ROPP-12.0.
Note:
See TracTickets
for help on using tickets.
Our approach is to compare the results of ROPP-x and ROPP-(x-1) by eye, for one compiler, at the beta review stage. Doing so automatically would imply rerunning the test folder using ROPP-(x-1) results as the new reference, and this would impose an intolerable computational load, since the full test folder takes a whole weekend to run - when it works.
Note that precise regression testing is performed, for one test per module, for our default compiler anyway - see the test folder results.
Changing ticket from 'enhancement' to 'task' as it needs to be repeated at each release. Promoting ticket to ROPP9.0 for the same reason.