Ticket #42 (assigned new feature request)
change summary for ciao's health
| Reported by: | jfran | Owned by: | jfran |
|---|---|---|---|
| Priority: | critical | Milestone: | |
| Component: | testing | Version: | |
| Keywords: | Cc: |
Description
The summary for Ciao's health may complicate as we add more tests for components. Currently those tasks are showed:
Clean, Configure, Build, Docs, Install, Tests, Uninstall
Although one could simply say that the task succeed or failed, the task Tests can contain a huge number of subtasks.
For every item we could define a tag:
% Tags for tasks (more than one are possible)
tasktag = obligatory
| compiler
| web
| tabling
| analysis
| performance
| ...
And for each task execution we may have those results:
% The status of the results
status = ok % passed
| failed % failed
| unverified % the output needs human verification
We may then present custom summaries that aggregates the results of everal tags, e.g.:
| O.S. | Obligatory, not Web | (Other) | Status |
| Ubuntu | 34 OK | 24 OK, 100 FAILED | OK |
| Debian | 30 OK, 4 VERIFY | 20 OK, 4 VERIFY, 100 FAILED | To verify |
| WinXP | 30 OK, 3 VERIFY, 1 FAILED | 20 OK, 104 FAILED | Failed |
Where Status should be defined depending on the requirements of the release (e.g. we may ignore all web related stuff).
How to verify the output
The method I use to verify the output (in trunk/ciao/optim_comp/ciaotool_modules/tests.sh) is to keep a copy of the last output that was ok. It already implements some predefined tasktags. It worked nicely for local usage, but misses some information like the revision number and the output needs to be verified for each machine.
The method in trunk/tests compares the output of the previous revision. The drawback of this method is that something wrong may happen and be forgotten later, once a new revision is committed.