Change bug notification email to point to new bug +subscriptions page to manage subscriptions

You can land the branch now - make it feature flagged :).

Then folk with the new flag get the new notice and folk with the old
get the old.

On May 18, 2011, at 9:52 PM, Robert Collins wrote:

> You can land the branch now - make it feature flagged :).
>
> Then folk with the new flag get the new notice and folk with the old
> get the old.

Our concern was that sending emails is done without requests. Setting up a context for feature flags to work will take time we don't have, and if we did set it up, checking feature flags naively would involve a whole new set of "potato-programming" queries for each recipient that might significantly slow down an important script. Again, we don't have time to deal with that right now.

Gary

Without any comment on scheduling this particular thing, I'll just
mention there are now flag scopes for incoming mail, and it would be
great (and hopefully not _too_ hard) to also apply them for outgoing
mail.

On Thu, May 19, 2011 at 10:59 PM, Gary Poster <email address hidden> wrote:
> Our concern was that sending emails is done without requests. Setting up
> a context for feature flags to work will take time we don't have, and if
> we did set it up, checking feature flags naively would involve a whole
> new set of "potato-programming" queries for each recipient that might
> significantly slow down an important script.  Again, we don't have time
> to deal with that right now.

Ah yes, you need an oracle for 'scopes applicable to user X'. Uhm.

So its fine not to do that, but here is how I would tackle it to avoid
those issues (and this makes the amount of work visible ):
 - setup a single feature controller without a request attached to it.
 - teach the controller how to discard its cached value for a flag
 - add to it a custom handler for 'inTeam:' (spelling?) - this handler
would change the inteam lookup from 'user in team' to 'get all the
members of the team named and do an in-python set lookup'
 - after checking the flag do 'controller.discard_cached_result('flagname')

-Rob

I guess we don't need to feature flag this now, and can just fix the issue directly?

Right, Jono, the order will be (1) give pertinent feature flags to everyone, (2) make sure nothing falls over, (3) land the branch we have that removes the feature flags and makes this change.

Fixed in db-stable r10611 <http://bazaar.launchpad.net/~launchpad-pqm/launchpad/db-stable/revision/10611>.