OK, I think we'll need to approach this from a different angle.

The way it's currently written, it will add a new filter that excludes everything typically excluded, but it also explicitly includes all of the new taskotron rules. I think it won't have any effect -- a taskotron message coming in will just be caught by and forwarded based on the existing filter that most messages hit -- the "Events on packages I own" filter.

Take a look at this older alembic upgrade script: https://github.com/fedora-infra/fmn.lib/blob/develop/alembic/versions/38c9c18d342e_add_new_mdapi_rule.py

It goes and finds everyone's existing "Events on packages I own" filter.. and then add more negated rules to it, making it more specific. (We needed to get everyone's preferences to ignore the very spammy new mdapi messages).

For taskotron.. we may need to take a step back for a moment and re-state what it is that we want to do. What kind of messages do we want people to get by default.. and what kind of messages do we not want people to get by default.

This is getting really complicated :-)

I believe we only want to notify people, when all of this is true:

• the tasktron result is related to their package
• the task completed is a release-critical task
• the task failed or changed state

We really wanted to add a new filter, because it is a combination of several rules and we can't easily filter out the complementary rules from "Events on packages that I own" filter. So the easiest solution seemed to be to create a separate Taskotron-specific filter (which people can easily tinker with, if they want, as a bonus).

You're right that currently it would have no effect, because all user-related taskotron messages would be allowed in by "Events on packages that I own" filter. So let's add a negated "all taskotron events" rule into that? That way Taskotron messages won't be processed by "Events on packages that I own" filter at all, and we will do all the processing in the new filter that is proposed. Does that work?

Yes @kparal. Good proposal. :)

@mkrizek, I'm going to close this since @kparal's proposal differs significantly. If you want to base further patches on this code -- that's cool -- just re-submit it. Thank you!

@mkrizek, I'm going to close this since @kparal's proposal differs significantly. If you want to base further patches on this code -- that's cool -- just re-submit it. Thank you!

@ralphbean I thought we were just going to add one rule (negate taskotron messages) to the existing packages filter in addition to using the existing patch. Did I misunderstand @kparal's proposal?