On Fri, Dec 6, 2019, 07:35 Johannes Lips <johannes.lips(a)gmail.com&gt; wrote: I agree, this is definitely a regression with newer bodhi versions (5.0+ I think). I also often get hit by bugs in updates that are either queued for stable, or already pushed. I would like to have the possibility to add such an information to an What's even worse, some updates skip updates-testing altogether when enough people give positive feedback on the koji builds fast enough. Then there's literally *no* opportunity for "normal" users of updates-testing to provide feedback in bodhi. Fabio

Il 06/12/19 07:34, Johannes Lips ha scritto: That was requested and discussed long time ago in https://github.com/fedora-infra/bodhi/issues/2050 The rationale behind this was about the nonsense to have users commenting on an already pushed update, since this cannot be undone. My opinion is that once an update is pushed to stable, new bugs should be reported in Bugzilla, not as comments in Bodhi. However there's an open discussion about restoring the previous behavior: https://github.com/fedora-infra/bodhi/issues/3748 The stable-karma and stable-days parameters can be set by the user in the web form or by CLI. By default stable-karma is set to 3, but it can be changed when creating the update. Mattia

Fabio Valentini wrote: IMHO, packages should just not be autopushed at all, no matter what threshold. An update can have +2 karma and be perfectly good to push, or it can have +10 karma and still have issues. It depends on what the users actually tested. So the maintainer should always read the freeform feedback text and decide based on that. (In fact, I would abolish the numeric karma entirely, provide only the freeform text field for feedback, and let the maintainer decide after reading it all. Humans will always make better decisions than naïve algorithms. But IMHO, even if the minimum karma requirement of +1 or 7 days wait time for non-critpath and +2 or 14 days wait time for critpath remains unchanged, it would still be an improvement to not allow automatic pushes of any kind.) Kevin Kofler

On Friday, 06 December 2019 at 10:57, Petr Pisar wrote: [...] I got hit by that as well when I deployed IPv6. GeoIP/geolite2 simply has wrong information about where your IP is located. Regards, Dominik -- Fedora https://getfedora.org | RPM Fusion http://rpmfusion.org There should be a science of discontent. People need hard times and oppression to develop psychic muscles. -- from "Collected Sayings of Muad'Dib" by the Princess Irulan

"Johannes Lips" <johannes.lips(a)gmail.com&gt; writes: We already have a tool for reporting issues and problems, and it's not bodhi. If there's a problem with an already pushed update, it needs to be in bugzilla - where it's actually discoverable - not in bodhi, where it will go nowhere. From the other side, as a maintainer, it is important to have as few channels for reporting issues as possible. Keeping everything in one place is important so nothing gets lost, and it all can be queried and triaged appropriately. Please do not use bodhi as a bug reporting tool. Thanks, --Robbie

On Fri, 2019-12-06 at 20:51 +0000, Johannes Lips wrote: FWIW, there is in fact an open Bodhi ticket for this: https://github.com/fedora-infra/bodhi/issues/3748 which references the earlier ticket that requested comments be disallowed on stable updates: https://github.com/fedora-infra/bodhi/issues/2050 -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net http://www.happyassassin.net

On Sat, Dec 7, 2019, 17:30 Mattia Verga via devel < devel(a)lists.fedoraproject.org&gt; wrote: I'm curious about this. I do get email notifications for all changes and comments for all my own bodhi updates plus for every update that I commented on. Have you turned this off, or do you mean to say you're going to ignore bodhi emails? It would be enough to let users comment on an update for, let's say, a week after it's pushed to stable. That way, immediate feedback isn't lost, but bugs that are found after that time period need to go to bugzilla. Fabio

On Sat, Dec 7, 2019 at 9:50 AM Fabio Valentini <decathorpe(a)gmail.com&gt; wrote: Speaking only for myself, yes I ignore bodhi emails. Why? I get huge amounts of email, so I've developed some survival strategies. One of them is to ignore sources of noise. Bodhi emails are mostly noise. Almost always, when bodhi sends me email, the email tells me something I already know. A build finished. Yes, I know. I saw the "fedpkg build" command complete. I didn't need even one email about it, much less multiple emails. If you want me to pay attention to bodhi emails, then the signal-to-noise ratio has to improve dramatically. It would be enough to let users comment on an update for, let's say, a week I agree with Mattia on this. Once an update has gone stable, there is no possibility of modifying it, so commenting on it further is not useful. File a bug please. Then I can create a new update to fix the problem, and a link to the update will be added to the bug. Here's the problem with allowing comments on stable updates: they're going to get lost in the shuffle. Those of us who maintain hundreds of packages need a fast, easy way to answer the question, "What outstanding problems do my packages have?" That way we can select what to work on next. I can go to bugzilla and see a list of open bugs. I can go to bodhi and see a list of updates in testing. Where, exactly, do I go to see a list of stable updates that have comments indicating more work needs to be done? Nowhere, which means I'm going to forget about those updates and the problems won't get fixed. Regards, -- Jerry James http://www.jamezone.org/

Il 07/12/19 19:32, Adam Williamson ha scritto: ...and that's exactly the behavior that the change tries to avoid. From the original RFE https://github.com/fedora-infra/bodhi/issues/2050 : Closing comments on a stable update will force users to refer to BZ, so that maintainers can focus on that channel. Instead, enabling users to comment on closed updates it probably means (50%?) that the same user will not open any bug in BZ, because we're all lazy, and "hey, I already posted a comment, I don't want to bother to write the same things on BZ"... In https://github.com/fedora-infra/bodhi/issues/3748#issuecomment-562867117 I proposed to give users a way to flag a problematic update with a link to a (previously opened) BZ ticket, so that 1) the first user that notice a problem with the update will be forced to open a ticket in BZ and 2) following users will be noticed that the update has been reported to have some problems and can CC to the discussion on the BZ ticket. Mattia

On Sat, 2019-12-07 at 10:46 -0700, Jerry James wrote: This part I agree with, actually - I don't monitor my Bodhi mail as closely as I should because of this issue. For example, take this update: https://bodhi.fedoraproject.org/updates/FEDORA-2019-81bcdfa3d5 Basically nothing of interest happened in its lifecycle. I created it, it got pushed to updates-testing, it reached the time threshold for being pushed stable, it was pushed stable. For that I get *five* emails notifying me of comments from bodhi itself: 1. "This update has been submitted for testing by adamwill." (yes, I know, that's *me*, I just *did* it) 2. "This update has been pushed to testing." (this is hardly news: it happens every time. Do I need an email to tell me about it?) 3. "This update can be pushed to stable now if the maintainer wishes" (maybe useful if I *don't* have autopush-for-time turned on, but is it any use if I *do*?) 4. "This update has been submitted for stable by bodhi." (obviously this arrived at exactly the same time as #3; there's really no need for both) 5. "This update has been pushed to stable." (again, this is hardly unexpected, do I really need yet another email to tell me?) Now, considering a fairly common workflow is to submit updates for multiple releases at a time - up to 6, if we have two stable releases, Branched, and you're submitted to three EPEL branches - this gets pretty silly: for 6 branches, I'm guaranteed to get *30* emails. Thus my Bodhi mail folder is a disaster area and I generally don't notice *interesting* things sent there, instead I spot them when I happen to go to the Bodhi web UI and look at updates there. Now, I could I guess do some fancy filtering for this client side; I could just try and filter out every mail that's only telling me about a comment from "bodhi" (although this isn't that easy because there is no convenient header to tell me this, it would have to be done by parsing the message content). But it might be nicer/more efficient to have some simple server-side choices here, like "don't send me mails for bodhi telling me about perfectly normal things happening", or something like that... Nit: Bodhi would not send you an email for this, because Bodhi has nothing to do with builds, only updates. At this point no update exists and there's nothing Bodhi could possibly email you about. -- Adam Williamson Fedora QA Community Monkey IRC: adamw | Twitter: AdamW_Fedora | XMPP: adamw AT happyassassin . net http://www.happyassassin.net