Merge lp:~lifeless/python-oops-tools/polish into lp:python-oops-tools
Status: | Merged | ||||
---|---|---|---|---|---|
Approved by: | Robert Collins | ||||
Approved revision: | 19 | ||||
Merged at revision: | 18 | ||||
Proposed branch: | lp:~lifeless/python-oops-tools/polish | ||||
Merge into: | lp:python-oops-tools | ||||
Diff against target: |
183 lines (+29/-20) 6 files modified
src/oopstools/NEWS.txt (+5/-0) src/oopstools/oops/models.py (+10/-10) src/oopstools/oops/templates/oops.html (+3/-1) src/oopstools/oops/test/oops.txt (+1/-1) src/oopstools/oops/test/pagetest.txt (+2/-1) src/oopstools/oops/test/test_dboopsloader.py (+8/-7) |
||||
To merge this branch: | bzr merge lp:~lifeless/python-oops-tools/polish | ||||
Related bugs: |
|
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
Jeroen T. Vermeulen (community) | Approve | ||
Review via email: mp+82365@code.launchpad.net |
Commit message
Show backtraces in OOPS report timelines.
Description of the change
Update the oops-tools model to show the new backtrace column in OOPS report timelines.
This is a bit crude - I'm making a somewhat horrid table more horrid (by making it wider). It also makes simple queries have a big gap between them.
OTOH there is a bit of a steep curve to get a js library in here due to our preferred tech not being the default in Django. I think we'll have a net benefit from this change without nice JS hiding, and I'm going to investigate JS hiding separately - e.g. Talk to francis about ROI, preferred choice etc etc.
This change can be easily backed out (its about 5 minutes to go from intent to deployed on our oops-tools instance a the moment) if we decide its a problem, so I'd like to get it up there and see.
Most of the changes have test coverage (in that tests broke when the change was made) except the actual template change itself. I've added a smoke test there (that the heading exists, which implies something about the body of the table), but its not truely satisfactory. OTOH I don't know what would be really satisfactory, given we're dealing with a pagetest here.
Thanks for your consideration :)
Is there no HTML test that would covers this at all? I do have a few concerns:
* Out of interest, any idea how costly it is to record, and later, render tracebacks for every event?
* What about HTML-escaping? What if the traceback contained a line like:
123, "</pre>", (a < b))
—wouldn't that upset the rendered HTML? Not likely to be a big problem for us, I suppose, but if we want this code to be reusable…