python-oops-wsgi doesn't record oops reports from Django's WSGI app
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Django |
New
|
Unknown
|
|||
python-oops-wsgi |
Invalid
|
High
|
Unassigned |
Bug Description
This may be better fixed in Django itself, but I thought I would report it here as well anyway since replacing wsgi-oops is one of the goals of this code base and it handles Django apps (albeit in a hacky way).
It seems that Django's WSGI application does not report the exception context to its container when it handles an error and displays an error page.
The way we're currently handling this is by (a) customising the Django Handler class we use to log the exception via the stdlib logging module, and (b) wsgi-oops collects the logged exceptions during the request.
A better solution would be for Django itself to do the right thing and not require this convoluted method of capturing the data.
Changed in python-oops-wsgi: | |
status: | New → Triaged |
importance: | Undecided → High |
Changed in django: | |
status: | Unknown → New |
Sadly I've found other things like web.py and flask also have this noddy approach. logging is in the standard library, so I think the glue for snarfing from logging output should be in python-oops itself.