/people/*/+translations times out for prolific translators

The URL mentioned is obsolete. I assume this is the updated one:

https://launchpad.net/people/menthos/+translations

That times out (today's OOPS-A208). Here's one that works:

https://launchpad.net/people/zkrynicki/+translations

Is this just a matter of batching or just truncating the results?

It would be good to sort by last activity and list it on the page.

Perhaps we could even make this use the Karma table, as this page and the Karma log both come from an "activity log" sort of angle.

It's not clear what the use cases for this page are. The ones I can think of are:

 * I'm looking at my own translation activity page to satisfy my ego
 * I'm looking at somebody else's translation activity page to see if somebody else is still active or has disappeared

I want this to stop oopsing.

But I can't seem to find a way out of this rathole. The problem is that POSubmission is just too massive, and using it to find out who has contributed to a template is too expensive. I think that a fix would be to cache this information in a TranslationContribution table, which would link people to pofiles (and which could contain some interesting information such as first/last_contribution_time. How do people feel about this? What are the gotchas?

A solution was discussed here:
https://lists.ubuntu.com/mailman/private/launchpad/2006-August/010391.html

The next step is now to Stuart create a cache table based on trigger mentioned in that thread.

The materialized view is done with tests and available at sftp://devpad.canonical.com/home/warthogs/archives/stub/launchpad/bug-2497

Patch that uses POFileTranslator is now up for review.

The patch is missing code to handle merging of the POFileTranslator table, which is necessary for merging.

bzr+ssh://devpad.canonical.com/code/stub/launchpad/bug-2497

*punt*

Landed in RF 4232.

The hard timeouts we were seeing are now gone. There are a few soft timeouts, but Matsubara has agreed to prioritize them when they become an issue again. Enjoy the new UI!

This is still an issue:

https://translations.beta.launchpad.net/~yannick-marchegay/+translations

After trying to get it three or four times, I got the page loaded so it's not as bad as before, but it's something we should fix

This seems to not time-out anymore, even if it takes a while to render the page (with Rosetta DB changes landed).

This has grown to time out again, and is increasingly timing out, in fact.

The Phase-2 schema optimization made a vast difference here, reducing timeouts by almost 90%.

More things we can do here:
 * Keep "translations to English" out of POFileTranslator, and remove the check for those from the slow query.
 * Upgrade the database server. There's a new optimization we'd be likely to benefit from with the above changes.
 * Index on (person, date_last_touched).

We now seem to be seeing this time out only on staging, thus reducing the importance to 'Medium'.

This doesn't seem to be a problem anymore.