Closed
Bug 502909
Opened 16 years ago
Closed 16 years ago
Stop using tail_ files in xpcshell: use do_register_cleanup() instead, in Toolkit Places
Categories
(Toolkit :: Places, defect)
Toolkit
Places
Tracking
()
VERIFIED
INVALID
mozilla1.9.3a1
People
(Reporter: sdwilsh, Unassigned)
References
()
Details
Attachments
(1 obsolete file)
Because tail_ files are ran right after run_test is called, the contents of those files may run before the test is actually done. This is very bad.
Attachment #387259 -
Flags: review?(mak77)
Comment 1•16 years ago
|
||
do you mean tails do not take in cound do_test_pending? sounds like a bug that should be fixed in tests harness, unless something prevents that.
| Reporter | ||
Comment 2•16 years ago
|
||
can't be fixed there because tests rely on it. tails run right after run_test is called.
Comment 3•16 years ago
|
||
it is still a bug if it does not take in count do_pending_test... any test relying on such a behavior would be crazy. The patch looks correct, but i'll wait for Ted to comment on the dependency.
| Reporter | ||
Comment 4•16 years ago
|
||
(In reply to comment #3)
> it is still a bug if it does not take in count do_pending_test... any test
> relying on such a behavior would be crazy. The patch looks correct, but i'll
> wait for Ted to comment on the dependency.
No, the intent was that it was supposed to run in order - head_*, test_*, tail_*. That's why I filed bug 502907 to add the cleanup method registration that allows us to run after all do_test_pending calls.
Comment 5•16 years ago
|
||
(In reply to comment #4)
> No, the intent was that it was supposed to run in order - head_*, test_*,
> tail_*.
indeed, the harness should wait for EVERYTHING one could put in test_* (included pending things) before executing tail_*, that's the bug i'm talking about.
| Reporter | ||
Comment 6•16 years ago
|
||
but it cannot because tests rely on that behavior. That's why ted and I agreed we need some function like this (and why I filed and patched bug 502907 which adds this API)
Comment 7•16 years ago
|
||
Comment on attachment 387259 [details] [diff] [review]
v1.0
at this point, looking at discussion in bug 502907, the do_register_cleanup should go in head_, and all tail_ files should be discarded. That will even simplify management since all common code for all tests will live in a single file.
Attachment #387259 -
Flags: review?(mak77) → review-
Comment 8•16 years ago
|
||
(In reply to comment #4)
> No, the intent was that it was supposed to run in order - head_*, test_*,
> tail_*.
I'm very confused by your comments here:
this issue was (fixed by) bug 384339 (a month ago), wasn't it?
(In reply to comment #7)
> the do_register_cleanup should go in head_,
> and all tail_ files should be discarded.
Indeed!
I'm "morphing" this bug per bug 503613.
Nit: you may want to add names to anonymous functions.
Severity: normal → trivial
Flags: in-testsuite-
Summary: Use do_register_cleanup in our tail files → Stop using tail_ files in xpcshell: use do_register_cleanup() instead, in Toolkit Places
| Reporter | ||
Comment 9•16 years ago
|
||
(In reply to comment #8)
> I'm very confused by your comments here:
> this issue was (fixed by) bug 384339 (a month ago), wasn't it?
Apparently, but it was landed with little fanfare, so I wasn't aware that it had changed.
Comment 10•16 years ago
|
||
(In reply to comment #9)
> Apparently, but it was landed with little fanfare,
Well, the "fanfare" was on MailNews bug 438922 (only),
as it seemed Core+Firefox tests were "fine" with how it (not-)worked.
> so I wasn't aware that it had changed.
I would have thought you would have noticed when you did the patch / filed the bug.
(Anyway.)
| Reporter | ||
Updated•16 years ago
|
Assignee: sdwilsh → nobody
Updated•16 years ago
|
Status: ASSIGNED → NEW
Comment 11•16 years ago
|
||
after bug 478718 we don't have anymore tail files. be happy.
Status: NEW → RESOLVED
Closed: 16 years ago
Resolution: --- → INVALID
Updated•16 years ago
|
Attachment #387259 -
Attachment is obsolete: true
Updated•16 years ago
|
Target Milestone: mozilla1.9.2a1 → mozilla1.9.3a1
You need to log in
before you can comment on or make changes to this bug.
Description
•