Code review comment for lp:~michael.nelson/ubuntu-recommender/944075-python-oops-wsgi

On Wed, 07 Mar 2012 15:15:22 -0000, Michael Nelson <email address hidden> wrote:
Non-text part: multipart/mixed
> Michael Nelson has proposed merging lp:~michael.nelson/ubuntu-recommender/944075-python-oops-wsgi into lp:ubuntu-recommender.
>
> Requested reviews:
> Canonical Consumer Applications Hackers (canonical-ca-hackers)
>
> For more details, see:
> https://code.launchpad.net/~michael.nelson/ubuntu-recommender/944075-python-oops-wsgi/+merge/96333
>
> Overview
> ========
> This branch just documents and adds tests to confirm that a configuration using oops-wsgi will work (and updates our schema to support it).

Hi Michael,

Thanks for working on this, it will be great to have this in use across
all our services.

> Turning this on in production will simply require updating the settings and django.wsgi in our config branch as documented here. But that can't be done until https://portal.admin.canonical.com/50529 is complete (unless we want to simply store the oopses locally, which could be done immediately with a config branch?)
>
> The new dependencies will need to be added via branches, as none of python-oops-wsgi, python-oops-dictconfig or python-oops_datedir_repo are available on lucid.

They are all packaged and available in lucid-cat though :-)
(You can use queries like "ssh people.canonical.com ~jamesw/cat-madison
python-oops-wsgi" to see what is in cat)

I'd be happy to work with you on getting the oops id available to the
500 handler, as it's a very useful thing, and something we will need in
many places.

> + class oops_wsgi(schema.Section):
> + oopses = schema.DictOption(default={
> + 'publishers': [{
> + 'type': 'datedir',
> + 'error_dir': 'oopses',
> + 'instance_id': 'dev',
> + }],
> + })

You could use the OopsOption from python-oops-dictconfig here.

> + def test_traceback_in_api_response(self):
> + # The API response should contain the traceback.

Is the API open for use by the public? Would we want to disclose
tracebacks in that case?

Thanks,

James