Meeting started at 2026-09-22 13:00:25 UTC

The Meeting name is 'Fedora Docs Team Meeting - 2026-09-22'

!hi

Hristo Marinov: Hristo Marinov (hricky) - he / him / his

!hi

Petr Bokoc: Petr Bokoč (pbokoc)

!hi

Eli Ridge: Eli Ridge (egret)

@pboy said he had some kind of a crisis at work so he can't come today

Justin Wheeler are you around?

Hmm. Anyway, let's get started

Everyone here knows about it already, but in case someone's reading the logs/minutes and isn't aware: We now have a separate channel for alerts - issues, PRs, build failures: #alerts-docs:fedoraproject.org

Be there or be square

That's it from me, does anyone have any other announcements?

None from me

All right

FWIW, I think this bot already left Docs room.

FWIW, I think this bot already left the Docs room.

Hah, has it? :-D

It might be waiting for our team ticket to conclude?

!hi

Jocelyn Gould: Jocelyn Gould (korora) - she / her / hers

So. This kinda sucks because Christian Strauf isn't here and he's the only -1, so we're not gonna solve this today. I really would like this to be an unanimous decision, if there's a hard no, I'd rather drop it since I don't think anyone is a hard yes, this isn't some kind of a "must-have" thing, and any gain from being included in TWIF doesn't counterbalance a potential loss of contributors. So we're gonna have to discuss this further. Christian also hasn't replied to the offer of excluding his name specifically, but I think his objection is to the bot as a whole, not just him specifically being tracked.

Hey Jocelyn Gould

I've also been wondering what would happen if we pointed the bot at the alerts channel instead and had it monitor stuff like new issues/PRs and comments.

And build failures. "This week in Fedora: pbokoc broke the build again"

I gave a -1 as well. Okay with the alerts channel though.

It's not a hard -1 one. I yield to the majority, no problem. Back to work. Sorry. :)

I don't know what benefit the bot would provide.

Right. Put it in the issue then :)

I don't feel comfy with an AI reading my every comment but...

Will do. Sorry, need to run.

"visibility" is the idea

I meant Eli :D

if someone wants to know they can... join the channel.

"Join the channel" would be my comment to people who want to know what's going on.

I did just before the meeting, so a bit late sorry

I did it just before the meeting, so a bit late sorry

Being lazy and having a bot summarise stuff isn't good style.

But that's just my 2 € cents.

Yeah. But they might not even know they want to know. The idea is, someone's scrolling through TWIF, and they get to our section, and they go "Oh, there's a docs team? I should check that out!"

Which, admittedly, is a very unlikely scenario, especially when our section is buried pretty deep in the report.

Wait, there's a docs team?

IMHO, the main concern is that what is posted to TWIF is not curated and comes directly from the LLM.

Yeah.

I'll throw some thoughts in the ticket.

Well, now that Eli voted (in the ticket) too, we're at +2/-2. That's... great :-D

I mean there's also a +1 from a "jamesthegiantpeach" but I don't know who that is

I'm leaning toward -1, just as an fyi

Discussing this further in the ticket is a good idea.

Okay, let's keep this in the issue comments for a bit longer and then we'll tell Aurelien.

Let's also keep discussing in the main channel.

Justin commented but maybe he is running late

Okay then! Let's switch topics before Christian gets written up at work :)

So first of all, there was a recent dicsussion - the site build failure notification script works great at the moment, but it gets really spammy in the alerts channel because it will notify on success, too, not just on failure. And there's no easy way to fix that. Earlier today, Hristo proposed two ways to deal with it:

Petr Bokoc, Justin Wheeler : Regarding timestamp notifications, I set the repo-level Nonbot webhook to only report failures, but the pings kept coming. It turns out they're actually coming from the docs org-level webhook, which is set to "All events." That setting bypasses per-event filtering entirely in Forgejo, so no combination of checkboxes on the repo-level webhook could have stopped it.

Two ways to fix it, and since this webhook covers every repo in the org, wanted to flag it here before touching it:

1. Switch the org webhook to "Custom Events" and replicate its current coverage minus Action Run Succeeded. Keeps it close to today's behavior, but "Custom Events" mode won't auto-include new event types Forgejo adds in the future — someone would need to remember to revisit it.

2. Uncheck both Action Run events (Success and Failure) on the org webhook, and let each repo's own webhook handle its own Actions/CI reporting (this repo's is already correctly scoped to failures-only). Org webhook stays focused on general activity — pushes, PRs, issues — across all repos.

Leaning toward option 2 since it keeps things simpler to reason about, but wanted input in case anyone's relying on Action pings from the org webhook specifically for another repo.

Thoughts?

Dammit. I forwarded that from our channel and it doesn't say who the OP is. For the record, that's Hristo's message.

Anyway, I think 2 is fine.

I didn't know that there was an alerts channel...

Yeah basically, we can disable Action Run events for all repos (org level) and then enable them in the repos that need them with the appropriate settings.

Hah! We've only had it for less than a week. Here: #alerts-docs:fedoraproject.org

I think #2 is fine.

But that needs to be done manually. I can try it.

Yay! more channels to join!

Okay, go for it

If option 2 keeps things more simple, that would seem reasonable

Also, I know about https://forge.fedoraproject.org/docs/docs-fp-o/pulls/58, I can merge it after the meeting. I'm not sure what exactly it means, though - does that mean the stage won't be monitored anymore? Or does the script on `prod` monitor both?

The script on prod monitor both and that is by design.

The script on `prod` monitor both and that is by design.

Okay, cool.

There must be only one schedule event on the main branch.

Ok then, merged

Is there anything else about scripts and CI we want to discuss right now?

Guess not. Which sucks, I was stalling hoping Justin would show up :-D

Since he's not here to tell us about docs captains updates, let's just go to open floor

Also btw, we should probably start a section on scripts in team docs and document what we have there.

I have been working on a git cheat sheet for team docs. Will likely need some input from more experienced people

I was going to suggest that. I can start writing a draft or at least open an issue on the topic.

At least the issue would be nice as a start.

Cool!

Oh, also, Christian had a report on the Beginner's Guide earlier, so let me get that into the minutes

I can also try to create a workflow which builds the Docs sites in the CI, if that would be useful.

Instead of OpenShift? That would be awesome, we could have logs visible to everyone

Not instead, but in parallel, along with previews. We can discuss the related topics in the Docs room.

I need to run, I'll catch the minutes after bit

Just to quicly jump in here we (fdwg) build some containers in our forge runner, feel free to copy our workflows, if they fit your usage!

Hey Vít Smolík. While you are here, and also to update everyone else - On ticket #26. Which is on collaborating writing. I have been unsuccessful in getting the cryptpad tool to work on openshift.

Could you please provide some links?

looking for them just now

https://forge.fedoraproject.org/fdwg/hatlas-fdwg-argocd/src/branch/main/.forgejo/workflows/build-superset-container.yaml

https://forge.fedoraproject.org/fdwg/forge_actions

hey! sorry, i completly lost track on that issue - i did some openshift stuff regarding public inbox, i can try and see if i can come up with a working cryptpad deployment

Yea put that one on the back burner as well. But anyway I am not sure if the resources are enough for cryptpad. It is a fairly complex thing to get running from what i can gather

Yea put that one on the back burner as well. But anyway I am not sure if the communishift default resources are enough for cryptpad. It is a fairly complex thing to get running from what i can gather

happy to answer questions / help you with this, if you want

Okay great. We're nearly at time so I'll end the meeting a little bit early

Thank you everyone for coming and I'll see you in #docs:fedoraproject.org! (And #alerts-docs:fedoraproject.org :P)