[Qgis-psc] again about the bug tracker
Vincent Picavet (ml)
vincent.ml at oslandia.com
Thu Oct 11 08:58:57 PDT 2018
Hello,
> This was raised to PSC since all the folks involved follow this list
> while the opposite is not true (PSC following qgis-dev). My intention
> was not to hide anything, I'm sorry for this.
>
> I have already read the page on the wiki a few times, and there is
> pretty much no change since Madeira.
> Some put a lot of energy to convince to investigate towards Gitlab and
> this relied on the production of an estimate of the cost of the migration.
Every single change as important and impactful as this one should have a
clear path, and an estimation of full costs and implications.
> This was the grant, which was rejected, but
> * why? because we want to stick to Redmine? because they prefer Github /
> not wasting energy? because voters prefer flashy features?
The latter is clearly the reason. And I do think this is the limit of
our current grant application program. It works well to decide for new
features, but not for ground-level, hard, not shiny but necessary work.
This has been seen multiple times with latest call for grant
applications. I can understand it from a user point of view, but I do
think this is the role of QGIS.Org to find a way to mitigate this.
Funding features is not hard, while funding "uninteresting" but
necessary work is. Bugfixing, bug triaging, PR Review, documentation,
infrastructure... My opinion is that QGIS.Org should concentrate on
these topics.
This is not the main issue here though.
> * it could have been ask to the PSC to sponsor this one (like many
> other projects: documentation, Python API, etc).
Two points here.
- I admit when the grant was rejected I have been clearly demotivated
and frustrated. I did not get any support except from a few people (
thanks to them ), and moved to more pressing things.
- The way anyone can ask the PSC to sponsor something in particular is
not clear to me. I did not even know it was possible except for
bugfixing efforts, before the specific documentation funding. I even
found it weird to see something funded outside grant applications, which
I believed was the only way (with bugfixing) to get something funded by
QGIS.org.
I really would like this to be written somewhere : who can ask, what
for, what conditions, etc.
> My point is that the road decided in Madeira (investigating the costs of
> moving to gitlab) is a fail at the moment. Of course, we can ask to be
> patient but this issue has been raised for at least 5 years. You say it
> needs time, I'm saying it needs energy. I'd love to some more
> pro-activity on the topic than just discussing here.
>
> My point that the current way is a fail might be wrong. At least, it's a
> personal point of view, I agree.
Yes, not mine at least. It is a work in progress, waiting for funding
and hard work. If you have one or the other available, I would be glad
to re-enter the game. Nothing is free and I rather prefer no engagement
into something than a failed engagement.
> In such case, we'll have to wait for another 6? months to get an
> estimate.
Not an estimate : a full migration plan. Again, we don't do pet-project
for such an important migration.
What would be an alternative doable faster ?
>To say it'll take roughly a year to move the CI (I don't think
> we can live with the code on Gitlab while the CI is on Travis/Github).
I don't see why we could not not live with code, issues, PR on gitlab
and CI working with a GitHub proxy.
> But in any case, things should move a bit faster.
Ready to move on, if we have resources available.
Regards,
Vincent
>
> Greetings,
>
> Denis
>
>
> Le mer. 10 oct. 2018 à 13:24, Vincent Picavet (ml)
> <vincent.ml at oslandia.com <mailto:vincent.ml at oslandia.com>> a écrit :
>
> Hi Denis, all,
>
> On 10/10/2018 15:32, Denis Rouzaud wrote:
> > The last issue about Redmine not sending mail makes me write again
> about
> > the topic.
> >
> > I'd like to propose to change direction from the decision which
> has been
> > taken in Madeira about moving to Gitlab.
>
> No decision has been taken in Madeira. I remember personally explaining
> that taking a migration decision should be backed by a full
> demonstration of capabilities, by clear explanations on advantages,
> drawbacks and implications.
> We should not do this kind of architecture change lightly.
>
> I have written all preliminary study results here :
> https://github.com/qgis/QGIS/wiki/QGIS-Platform-migration-plan
>
>
>
> > I'ts been roughly half a year and nothing moved except for a declined
> > grant proposal.
>
> Yes, for this kind of action, we need time, we need funding. I applied
> for a grant, and it has been rejected. I would be glad to resume the
> work on this subject, but it needs interest and funding, or it will not
> move forward.
>
> > I have heard something from Steven Feldman at the FOSS4G which rang a
> > bell. I don't recall the exact formulation nor my phrasing is as
> precise
> > but he advised to be pragmatic and to avoid losing too much energy on
> > ethical or not-strictly-related-to-the-topic issues...and to me, we're
> > looking at something (Gitlab) which represents weeks of
> development just
> > for the CI and which barely bring anything valuable over Github
> while we
> > stick to a non satisfying solution (Redmine).
>
> Please Denis, at least read the mail archives and the work already
> achieved before posting this kind of false statement :
> https://github.com/qgis/QGIS/wiki/QGIS-Platform-migration-plan
>
> It is true that CI has to be tested, but keeping a read-only GH copy
> should be enough to get Travis working, and give time for full migration
> to GitLab.
>
> On the Github side, nothing has been done to prove that a migration is
> even doable without problem.
>
> > I would not deny that ethical is important...but what/who are
> working for?
>
> This is your appreciation of things, and it may largely differ from
> person to person. Trying to push this opinion directly to PSC although
> we have already discussed these topic more broadly does not seem the
> right thing to do.
>
> Anyway, if we want to go forward with this topic, then the initial
> proposal of the grant application is still valid. I am confident that
> with budget we can finish up the migration study and reach a state where
> we proved that it is a good way to go.
>
> Any concrete help is welcome too on this subject.
>
> One other open question also is the cost of hosting and maintaining the
> solution, but this is more or less the same as for Redmine.
>
> Best regards,
>
> Vincent
>
> >
> > Best wishes,
> > Denis
> >
> > --
> >
> > Denis Rouzaud
> > denis at opengis.ch <mailto:denis at opengis.ch>
> <mailto:denis at opengis.ch <mailto:denis at opengis.ch>>
> > +41 76 370 21 22 <tel:+41%2076%20370%2021%2022>
> >
> >
> >
> >
> > _______________________________________________
> > Qgis-psc mailing list
> > Qgis-psc at lists.osgeo.org <mailto:Qgis-psc at lists.osgeo.org>
> > https://lists.osgeo.org/mailman/listinfo/qgis-psc
> >
>
> --
>
> Denis Rouzaud
> denis at opengis.ch <mailto:denis at opengis.ch>
> +41 76 370 21 22
>
>
More information about the Qgis-psc
mailing list