Proposed F19 Feature: systemd features

["Jóhann B. Guðmundsson"]

On 02/02/2013 02:28 AM, Adam Williamson wrote:
> On Sat, 2013-02-02 at 01:25 +0000, "Jóhann B. Guðmundsson" wrote:
>
>> Feeling happy in the Red Hat position they invented for you in QA.
>> Feeling a big man now? Challenged accepted big man you have in your rein
>> of error effectively killed 2 thriving process in the QA community. Do
>> you really want to head down this road with me? Go ahead big  man make
>> my day!
> Well, no, not really. We're not having a fight, here. At least, I'm not
> trying to. I'm trying to discuss process. You appear to think this is a
> battle for domination, but that's not how it looks from where I'm
> sitting. I'm sorry if it came up that way, I'm not trying to slap you
> down or demean you, I think we've just fallen prey to a bit of process
> confusion here.
>
> So, can we back up a bit? I've just read back the whole way through the
> two threads, and it turns out to be a bit of a mess. I think the problem
> is different people have different perceptions of what's going on.
>
> So in the 01-16 FESCo meeting - right at the end, in open floor - there
> was a short conversation about cron job conversion:
>
> 18:56:15 <Viking-Ice> on an different subject does it make sense to
> migrate all those cron scripts to systemd timer units?
> 18:56:25 <mmaslano> what?
> 18:56:25 <Viking-Ice> all shipped cron scripts I mean
> 18:56:40 <mitr> Viking-Ice: what end-user benefit would all that work
> bring?
> 18:57:10 * nirik has no opinion on it without looking at it in more
> detail.
> 18:57:20 <mmaslano> Viking-Ice: as a maintainer, I don't plan to do it,
> because I don't see any user improvement
> 18:57:37 <mmaslano> is there any?
> 18:58:11 <nirik> lets discuss that proposal after it's had discussion on
> list or detailed proposal sent to us?
> 18:58:22 <Viking-Ice> mitr, fewer packages in install have not looked at
> it further it's just recent that systemd gained "calender" support
> 18:58:40 <Viking-Ice> for those timer units
> 18:59:31 <Viking-Ice> I'll compile a list of packages that ship cron
> jobs and look what benefits it might bring
> 18:59:42 <mmaslano> Viking-Ice: could you mention in your proposal what
> it brings to users?
> 18:59:44 <mmaslano> ok thanks
>
> This wasn't widely advertised at the time, it's not mentioned in the
> 01-16 meeting minutes (no #info or #action lines were given). So anyone
> who wasn't at the meeting probably never saw it. The discussion does not
> relate to the actual feature that was proposed - as we agree, that does
> not cover cron job migration. And so far as I can tell you never brought
> any kind of formal proposal to FESCo: there is nothing in the 01-23 or
> 01-30 logs on this topic, and you have not proposed a feature for it.

No because I was looking into it as I have been trying to state this 
whole time.

> There was then the thread "Start of systemd timers after install/update
> of a package". You mentioned that you had looked into the migration
> possibilities and come to a list of 38 cron jobs in service related
> packages.

Yes out of 99

> There was a bit of follow-up discussion between you and
> Marcela; other principals in the current discussion weren't involved in
> the thread at this point (Bill posted to it early, but never *after* you
> mentioned your list of 38 packages).

Bill showed up on the systemd channel looking for a glorified solution 
in the form of generators but mmaslano had already at that time shown 
her concerns about potential administrative conflicts as in half in cron 
job half in timer units...
> Then we come to this thread, which was related to a specific feature:
> https://fedoraproject.org/wiki/Features/SystemdCalendarTimers . The
> migration question was brought up early and quite vaguely by Marcela,
> not clearly in direct relation to the feature but more as a 'general
> concern'. You replied and clarified that the feature doesn't actually
> cover the cron job migration, and re-stated that you were working on a
> cron job migration proposal.

Yes timer units have been there from more or less the start and they 
just gained calendar support which is totally irrelevant to my research. 
Tim wanted even to put in on our ( QA ) meeting logs that he showed 
concerned over this failing to understand that all that has happened is 
that you can specificy calendar days dates in timer units which he then 
retracted from the meeting logs once I had explained it to him that, 
that feature had nothing to do with anykind of migration.

>
> Then we fast-forward a week or so, and Bill replied to Marcela and
> provided his own proposal for how to decide what cron jobs to migrate.
>
> Now at this point I can kinda see how to you it looked like Bill was
> 'butting in on' your work, as now I've gone and done the thread
> archaeology and dragged up three week old FESCo logs from the archives,
> I can see that this has been kind of an 'ongoing thing' for you. But it
> doesn't quite have that visibility for anyone else - the FESCo
> discussion wasn't in the minutes and there's no public draft proposal,
> and the references in the "Start of systemd timers after install/update
> of a package" thread are somewhat vague and it's not clear Bill even saw
> them. Bill was present for the open floor section of the 01-16 meeting
> (I just checked my personal logs, as the public ones don't show
> joins/parts), but he didn't say anything - he may not really have
> followed the conversation. So Bill may not actually have been fully
> aware this was something you were working on.

Well I just hope so that an elected member of fesco is paying attention 
on the meeting they have been elected for, for the whole community if 
not he should resign his position from my pov..

>
> This was where the thread started going seriously downhill - in response
> to three different people you stated more and more strongly that you
> were working on the cron job migration thing and Bill was 'butting in'.
> Again, I can see how it looked like that to you, but I think to the
> others it didn't - I'm guessing the others, like me, hadn't seen all the
> previous stuff. It looked more like Bill suggested a proposal and you
> were claiming more or less by fiat that this was your area and he should
> butt out, and that's what got people's backs up.

As soon as I saw that Bill had started to butt in then I asked Rahul to 
stay out of it because at that time it had gone to "personal", mmaslano 
had already realized the complication and shown her concern of shipping 
mixed as in half in timer units half in cron jobs.

>
> So now I've dug through the background I can see there's two sides here
> and I'm sorry for not catching that you actually had been working on
> this and you had some kind of justification for feeling the way you did.
> But can you accept that it looked different to other people, and your
> response was kind of overheated? When you start telling people to 'stay
> out of this' and accusing people of 'butting into your work' it kinda
> escalates the whole thing.

No not really when it comes to Bill. Sorry no empathy there what he did 
to me during the F15 cycle is unforgivable. You may hold him high on 
some petostole I do not and never will. Not you nor Kay or anyone else 
will convince me other wise. I have seen his true nature and I can deal 
with it. I have dealt with people like him before I can do it again but 
I will never expose that to people I'm working with hence the dissembles 
of the systemd migration crew.

>
> So, sorry again for not catching how this looked from your angle, can we
> agree that it was the same problem on both sides and just move on now?

No the damage is done.

> FWIW, it does look to me from the 01-16 logs like you and FESCo had
> agreed that you'd come up with a proposal for the cron job migration
> stuff, so it seems reasonable to ask Bill to work together with you on
> that, or if it's impossible for you two to work together, maybe you
> could both come up with proposals and FESCo could try to synthesize
> them.
After I refuced to send all our reporters to "upstream Gnome" to file 
report, after the systemd migration treatment he gave me, after I filed 
a proposal for the "rsyslog/journal", even packaging proposals so we in 
QA would have a head start dealing with potential breakage from it and 
now for the cron jobs the might get migrated. no not going to happen.

> Still, so long as you hadn't actually filed some kind of formal
> proposal or feature with FESCo, it's hard for anyone to know this was an
> area you were 'officially working on'.

There is nothing to work on until I ( and fesco ) know what are dealing 
with hence the research "first" then the outcome ( which might be 
migration proposal or not ).

You actually have to install and look at the cron jobs to see if they 
are worth migrating to timer units.

JBG