[mapserver-dev] important news item: Travis CI has changed its pricing model

[Steve Lime]

Jeff, you're right, we should scare up a new PSC meeting to discuss. We're
sitting on $2,500 that needs to be spent so covering a year to come up with
a transition plan - if necessary - certainly seems reasonable. --Steve

On Mon, Nov 2, 2020 at 12:23 PM Jeff McKenna <jmckenna at gatewaygeomatics.com>
wrote:

> On 2020-11-02 2:05 p.m., Even Rouault wrote:
> > On lundi 2 novembre 2020 11:21:29 CET Jeff McKenna wrote:
> >> Hi devs,
> >>
> >> It's important to point out that as of today Travis CI has changed its
> >> pricing model, so that Open Source projects get a free block of 1000
> >> minutes of build time (only about 16 hours), and will have to re-apply
> >> to get further blocks of time.   Related blog:
> >> https://blog.travis-ci.com/2020-11-02-travis-ci-new-billing
> >>
> >> To give you an example, I bet I used about 10 hours of build time on
> >> MapServer last week myself, yikes.
> >>
> >> This change also means a move from travis-ci.org to travis-ci.com for
> >> this, for example:
> >>
> >>     - see our list of repositories at
> https://travis-ci.org/github/MapServer
> >>     - see how none of our repositories are setup yet at
> >> https://travis-ci.com/github/MapServer
> >>
> >> I have started the process and now have 1000 minutes for my own account
> >> on .com but am looking at how to set this up for MapServer and all of
> >> our repositories.
> >>
> >> The knee-jerk reaction to this will be "we need to move to Github
> >> Actions/workflows for builds right away", but I think we should examine
> >> and discuss all options together before any quick decisions are made for
> >> the MapServer 'organization' (Github's term referring to all of
> >> MapServer's repositories).
> >>
> >
> > A theorical alternative would be to see if Travis-CI would accept the
> > MapServer organization to share the paid premium plan of the OSGeo
> > organization, that is mostly used by GDAL, PROJ and GRASS. But I'm not
> > confident they would be willing. And I'm not sure either of the future
> of the
> > current OSGeo premium plan. I wrote to Travis support to see what their
> > billing changes would impact and their answer is far from being crystal
> > clear... It seems we'd have to choose between concurrency plans or usage
> based
> > plans as defined in https://travis-ci.com/account/plan , but their
> maximum
> > concurrency plans is 5 instead of the 10 we have currently for OSGeo (and
> > originally we had 13 when we started on the paid plan). And the usage
> based
> > plan doesn't seem appropriate for open source projects and pull requests
> > mechaism as it implies to limit/control the number of users who submit
> builds
> >
>
> Yes, and it could make sense to actually use our MapServer budget to
> purchase one of the monthly plans for Travis CI builds (if you want my
> opinion this could be very worth it, as I'm all for paying for something
> to keep our architecture stable, instead of jumping to the next
> free-turning-proprietary-turning-paid service).
>
> For example: even the lowest/fewest concurrent job plans on
> travis-ci.com looks like a great option for the MapServer organization:
>
>   - 1 concurrent job = $69/month = $828 yearly
>   - 5 concurrent jobs = $249/month = $2,988 yearly
>
> I think spending the $828 out of the MapServer budget each year to have
> a dedicated (and stable) build environment just for the MapServer
> organization is a good plan.  Only 1 concurrent job means that
> technically we can have some developers waiting sometimes, but as most
> times it is me doing the build, and recently the OSS/free Travis CI plan
> now takes ~1 hour to wait for our free build turn before the virtual
> machine spins-up, I think the dedicated instance would be a big
> improvement and even *less* wait times than the current free/OSS plan.
>
> -jeff
>
>
>
>
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/mapserver-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/mapserver-dev/attachments/20201104/660dc9b3/attachment.html>