Some valgrind testing on the RQ
Bug #1185221 reported by
Chris Hillery
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| Zorba |
New
|
High
|
Chris Hillery | ||
Bug Description
Somehow, some way, we need to run some subset of our tests with valgrind on the RQ, to prevent the introduction of memory errors.
It is not clear to me how to do this as I do not think we have anything like a baseline set of valgrind reports - it is certainly the case that not everything runs valgrind-clean, and some of those reports are either unavoidable or are in dependent libraries.
I'm also not sure how to determine a reasonable subset of test cases to run which will offer decent coverage without making the RQ take hours to run.
Revision history for this message
| Chris Hillery (ceejatec) wrote | #1 |
| Changed in zorba: | |
| assignee: | nobody → Markos Zaharioudakis (markos-za) |
| milestone: | none → 3.0 |
| importance: | Undecided → High |
| Changed in zorba: | |
| assignee: | Markos Zaharioudakis (markos-za) → Chris Hillery (ceejatec) |
| Changed in zorba: | |
| milestone: | 3.0 → 3.1 |
To post a comment you must log in.
It may be that we *also* want to run a broader set of tests with valgrind (since running everything that way on the RQ would make it take hours). In that case, we should IMHO do this nightly or weekly, rather than at the end of the release cycle, and any reports brought up by such testing should be fixed with highest priority. However, for this to be useful, we also would need some kind of baseline valgrind report that we deem acceptable, and only report regressions.