On Fri, Dec 21, 2018 at 11:37 AM Lionel Landwerlin
<lionel.g.landwer...@intel.com> wrote:
>
> On 19/12/2018 17:19, Kenneth Graunke wrote:
> > On Wednesday, December 19, 2018 8:10:49 AM PST Eric Engestrom wrote:
> >> The point of this patch is for people to not be able to ignore this, and
> >> have to become aware of the existence of Meson and our intention to
> >> remove autotools, whenever this may happen.
> >>
> >> Some people have already noticed, probably because they keep up with
> >> mesa-dev@, but the point is that some people haven't.
> >>
> >> Without this patch:
> >> - people will continue being unaware, until
> >> - autotools gets removed, and these people are going to try their
> >>    use-cases, and probably find issues
> >> - issues will get fixed, hopefully quickly, but blocking them completely
> >>    in the mean time
> >>
> >> With this patch:
> >> - people will *have to notice* that something's changing
> >> - they'll choose to either ignore it, use the flag, and *choose* to go
> >>    with the option above
> >> - or they can choose to try meson, get any issues fixed *while still
> >>    being able to work with their old setup in the mean time*
> >>
> >> This last bit is the entire point of this patch, so let me repeat it:
> >> people will report meson issues while there's still time to use their
> >> old autotools setup.
> >>
> >> I'm definitely in favour:
> >> Acked-by: Eric Engestrom <eric.engest...@intel.com>
> > Eric's reasoning here makes sense.  In case there is anyone out there
> > who has been putting off trying Meson, this will properly exhort them
> > to try it, and identify any problems, yet still let them continue using
> > their old system for a little while longer.
> >
> > Acked-by: Kenneth Graunke <kenn...@whitecape.org>
> >
> > I've been using Meson for a long time, and it is dramatically better to
> > work with - both to write, and to use.  It is a little bit different,
> > and that took some getting used to.  But it was absolutely worth it.
> >
> > We need a way to make progress here.  Matt's concern is very valid -
> > people can simply keep moving the goal posts.  And that's incredibly
> > frustrating for the people doing the work.
> >
> > Given that we've had the Meson build system for over a year, I think
> > time is up.  Marek once received review feedback to a patch a year after
> > he wrote it, and he responded with something like "I've moved on, feel
> > free to write a patch."  I think that applies here.  The Meson build has
> > had many contributors at this point.  If it's not good enough, feel free
> > to help fix it.  It's worth noting that it is good enough for the major
> > distros, which have switched, and a large portion of the community uses
> > it exclusively.  Dylan has done a ton of work to upstream Meson to
> > address our needs and improve the usability of the project.  The rest
> > of the Meson developers have been very responsible as well.  We have
> > every reason to believe it will keep getting better.
> >
> > We can't keep kicking the autotools can down the road indefinitely.
> > We need a forcing function like this patch, then eventual removal.
> > Expecting new contributors to write 4 build systems is just...bad.
> > I'm hopeful that we can get down to two (Meson and Android) soon.
> >
> > For what it's worth, I'm hoping to merge Iris soon, and it does not have
> > an autotools build system, nor do I intend to write one.  At that point,
> > you won't be able to build the new Intel OpenGL drivers with autotools.
> >
> > --Ken
>
> I very much agree with Ken here.
> Autotools has been a pain to deal with especially since we had to
> maintain 2 build systems.
> We already have part of our tools that do not get built with Autotools.
> I think the main reason being that debugging Meson builds is a lot
> faster that Autotools ones.
> So I would be happy if we could send a signal that this will go soon and
> strongly recommend users to tell us what isn't working.
>
> Just finally adding that pretty big projects like systemd (only uses
> Meson), GLib (both Meson/Autotools, but Meson default) and a bunch of
> other Gnome stuff is now being built with Meson in major distros (just
> checked Debian which also builds Mesa using Meson), so it seems to me at
> least a significant part of the community won't be very surprised.
> And this patch is the call for people who haven't noticed yet.
>
> Acked-by: Lionel Landwerlin <lionel.g.landwer...@intel.com>
>

I also agree that maintaining 2 build systems (well, actually 3
because android) is major pita, and meson is much nicer to deal with
than autotools.  I think the approach that this patch is taking, to a
little more actively push people who haven't tried meson to give it a
try and speak up if something is missing, is extremely reasonable and
far more conservative than what a lot of other projects have done.

Acked-by: Rob Clark <robdcl...@gmail.com>

> -
> Lionel
> _______________________________________________
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to