Launchpad shouldn't encourage users to file bugs when an OOPS occurs

[Jani Monoses]

Now that we have high quality tools for dealing with OOPSes (James H's summaries), QA processes for dealing with OOPSes (https://wiki.launchpad.canonical.com/LaunchpadBugTriage), and a QA team who are looking at OOPSes regularly, it's no longer necessary to rely on users to file bugs about OOPSes.

When the support tracker has been improved, we could perhaps add a link to the support tracker to the OOPS page.

[Andrew Bennetts (spiv)]

I think this is a pretty good idea, although there are times when we will want to contact the original reporter for more information, and at the same time I think it would be rude to automatically subscribe people to bugs they didn't file. Launchpad developers can get that information out of the oops report, if the user is logged in.

If the user isn't logged in, we'd probably want to update the error page to ask the user for an email address where we can contact them for more information if we need it.

Also, there's the considerable technical problem of dealing with what happens if the auto-filing procedure has a bug and fails?

Nothing insurmountable, but some thought required.

[Dafydd Harries (daf)]

The problems I can see with automatic filing of bugs:

 - it won't work for people who are not logged in (we require somebody to be logged in to file bugs)
 - I suspect that many duplicates would be generated
 - many OOPSes are not bugs -- e.g. many 404s are perfectly valid
 - the sheer number of bugs that would be filed -- we're currently getting > 2500 OOPSes a day according to James H's latest report

I would favour a solution that makes the filing of bugs related to OOPses easier by filling in the title/description with a template.

[Jani Monoses (jani)]

>
> - it won't work for people who are not logged in (we require somebody to
> be logged in to file bugs)

is it hard to fall back to a default reporter in this case ?

- I suspect that many duplicates would be generated

most of those would be duplicates even if a human filed them. But using the
context which may be known by LP these can be categorized so that it's easy
to see which are probably duplicates.

- many OOPSes are not bugs -- e.g. many 404s are perfectly valid

when the error message does not require the user to file a bug it should not
autofile bugs either.
in case of a 404 can LP detect whether it was referenced from a valid LP
page and only then consider it a bug?

- the sheer number of bugs that would be filed -- we're currently getting >
> 2500 OOPSes a day according to James H's latest report

again that is only smaller otherwise if you hope people will not file bugs
on each oops. Which they tend to do ;)

I would favour a solution that makes the filing of bugs related to
> OOPses easier by filling in the title/description with a template.
>

that too is a step in the right direction. The user then should specify what
she was trying to do so the process is reduced to
editing a box then pressing submit.

[Dafydd Harries (daf)]

We require people to be logged in to submit bug reports so that we can communicate with them later. Using a default reporter would defeat this.

It's true we could exclude 404 bugs from being automatically reported.

Our current plan is to change the OOPS page so that it doesn't encourage users to file bug reports. We now have better tools for helping us to monitor OOPSes, so there is much less of a danger of OOPSes going unnoticed if users don't file bugs. Are you happy with this solution?

[Jani Monoses (jani)]

>
>
> Our current plan is to change the OOPS page so that it doesn't encourage
> users to file bug reports. We now have better tools for helping us to
> monitor OOPSes, so there is much less of a danger of OOPSes going
> unnoticed if users don't file bugs. Are you happy with this solution?

Sounds good, thanks.