[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Alioth replacement and pkg-perl



Hi,

I really wanted to answer sooner, but work and life got in the way.

On 24/08/2017 18:30, gregor herrmann wrote:
> On Thu, 24 Aug 2017 00:32:20 +0200, Alex Muntada wrote:
> 
>>> Will try to attend, without any garantee as I'll be doing the
>>> Alioth Sprint thing.

And indeed I was exhausted and had to deal with a domestical
emergency... Sorry for not even popping by on IRC.

>> Thanks for your work on the Alioth sprint last weekend. Even
>> though the sprint is over, I'll follow up with some details
>> regarding the questions you raised below.
> 
> Thanks for this summary!
> I'll just add some details.
>  
>> AFAIK pkg-perl most used service in alioth is git, plus remote
>> hooks that send notificacions to IRC and lists, and another hook
>> that builds the HTML web pages from POD files in website repo.
> 
> And PET.
>  
>> KGB instances run on dmn's, gregoa's and Tincho's servers.
> 
> Right, and the client is an Alioth, triggered by our hook script.

Hook scripts will definitely have to work differently. I don't think the
local hooks of gitlab will get implemented, but maybe there could be an
exception for KGB, or we could doo something with the CI runners.
The usual way to do that in gitlab is to use the webhooks, but it means
we need a rest API of some kind to receive them.


>> PET seems to run on petrova.debian.org, so it's not an alioth
>> service though it may depend on some alioth magic like repos
>> traversal, running local scripts, etc. that send changes to PET.
> 
> Right, it's also triggered by our hook script (and a cronjob with
> "update-all" just to make sure).

Same as above, then.

<skipping more alioth stuff for which I have no solution in mind>

>> The annual ping reminder for pkg-perl members is run in alioth
>> via pkg-ruby-extras' find-inactive-contributors, plus some
>> additional steps for pkg-perl. We'll need to figure out how to
>> do the same for git.d.o replacement.
> 
> And I guess we have some time for it, as the new pkg-perl team on
> salsa.d.n will start out with 0 members, if I understand the plans
> correctly.

Exactly.

>  
>> There's a proof of concept pkg-perl manual in the website repo
>> that is build from POD files via Sphinx and then rsync-ed to
>> alioth manually. Since Sphinx is getting traction in other
>> Debian teams too, I expect that some kind of service will be
>> setup eventually that we could use too.
> 
> Right.

When I tested the gitlab pages with current pkg-perl code, I just ran
the sphinx stuff locally and pushed them via git.
That said, formorer said to me on IRC that supporting sphinx or other
lightweight html generation directly on the server wouldn't be a problem.

>> That's all I can remember so far. 
> 
> There are also scripts/cronjobs to produce the 2 files in
> https://pkg-perl.alioth.debian.org/qa/
> but I guess we can live without them :)
> 
>> I guess we'll need to draft
>> some plans for things that are going to break when alioth is
>> put in read-only mode and finally shut down.
> 
> I think it's good to collect things that we need (as we do right now
> :)); for most issues I hope there will be general solutions or howtos
> etc. as many projects are affected.
> 
> 
> Maybe we should start a gobby doc or wiki page to list our tasks?

Yes.

I was hopping for the official report of the Spring last weekend, so
that we would know for sure where we stand currently, but it's still
pending. We probably should start writing stuff and can amend it later :)



Cheers,

-- 
nodens


Reply to: