Merge lp:~mbp/bzr/626679-strace into lp:bzr
| Status: | Merged | ||||
|---|---|---|---|---|---|
| Approved by: | Robert Collins on 2010-08-31 | ||||
| Approved revision: | 5399 | ||||
| Merged at revision: | 5402 | ||||
| Proposed branch: | lp:~mbp/bzr/626679-strace | ||||
| Merge into: | lp:bzr | ||||
| Diff against target: |
184 lines (+54/-28) 3 files modified
NEWS (+4/-0) bzrlib/strace.py (+22/-3) bzrlib/tests/test_strace.py (+28/-25) |
||||
| To merge this branch: | bzr merge lp:~mbp/bzr/626679-strace | ||||
| Related bugs: |
|
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| Robert Collins (community) | 2010-08-31 | Approve on 2010-08-31 | |
|
Review via email:
|
|||
Commit Message
disable strace tests broken under maverick
Description of the Change
Ubuntu recently changed to stop users tracing their own processes by default as a security measure, which means that 'bzr selftest -s bt.test_strace' fails on maverick.
Unfortunately strace doesn't give you a clear error code to tell you that you hit this.
We don't actually use strace in any other tests at the moment, but it has been useful in the past for ad-hoc performance investigation.
| Martin Pool (mbp) wrote : | # |
sent to pqm by email
| Andrew Bennetts (spiv) wrote : | # |
PQM seems to have hung while processing this. The circumstantial evidence from spm is that it was probably executing the strace tests at the time: there was an strace subprocess, and killing it caused another strace subprocess to be spawned. Hopefully some informative tracebacks/errors are on their way back to Martin from PQM.
| Martin Pool (mbp) wrote : | # |
Thanks.
That actually pushes me a bit more towards deleting/disabling the
tests, but leaving the facility. It's useful to have it, but it seems
silly to have test hangs for something that's not actively used other
than in the tests.
--
Martin
| Andrew Bennetts (spiv) wrote : | # |
Martin Pool wrote:
> Thanks.
>
> That actually pushes me a bit more towards deleting/disabling the
> tests, but leaving the facility. It's useful to have it, but it seems
> silly to have test hangs for something that's not actively used other
> than in the tests.
I think that's a reasonable compromise. Maybe leave the tests there but
skipped? It doesn't seem like something worth spending large amounts of
time on. The next person to work on the strace helpers can look at
fixing their tests if they like.
- 5400. By Martin Pool on 2010-09-01
-
Remove TestStraceFeature that just duplicates the code it tests
- 5401. By Martin Pool on 2010-09-01
-
Just skip the strace tests
| Martin Pool (mbp) wrote : | # |
sent to pqm by email

+1