[cfe-dev] [llvm-dev] GN build roundtable summary; adding GN build files to the repo

Tue Nov 6 03:10:38 PST 2018

The value in having them somewhere in-tree is that it's easier for people
collaborate on these files, and it's way lower setup overhead if someone
wants to try it out. If people prefer llvm/util over llvm/experimental,
that's fine with me.

There would only be a single directory that will contain build files for
all of llvm, clang, lld, etc. The build files would be in
llvm/whatever/gn/tree/{llvm,clang,lld,...}/... ("tree" can be any name).
That's not the most natural way to use gn, but I agree that having BUILD.gn
files scattered all over the tree could make it look like gn is supported
at some level.

If adding the files in an isolated directory still ends up being confusing
in practice, we can move them out at that point. But I'd like to start with
the simpler way and see if that's good enough.

On Mon, Nov 5, 2018 at 8:22 PM Justin Bogner <mail at justinbogner.com> wrote:

> Nico Weber via llvm-dev <llvm-dev at lists.llvm.org> writes:
> > If I read this correctly, there isn't much opposition to landing the gn
> > files as long as it's very clear that regular devs aren't supposed to
> > update them and that it's clear that they're experimental
> >
> > The main concerns I've heard so far:
> >
> > - Having two build systems is confusing. I can see this, but I think
> > putting the gn files below llvm/experimental/gn (instead of right into
> the
> > source, like I currently have) and updating GN.rst to explicitly say
> > "Reviewers should not ask patch authors to update gn files in addition to
> > the cmake files" will address this.
>
> I haven't been following this discussion and I don't really know
> anything about GN, but I'm not very convinced by this.
>
> I'm opposed to adding a second build system in general. We all remember
> having two build systems before and it's a pain keep track of as well as
> a documentation nightmare in telling people yet more ways to do
> things. Our docs are already confusing enough given that we currently
> support two completely different layouts of source directories for
> cmake.
>
> I suppose putting it somewhere under utils/ as a kind of tool for people
> who want to use it like we do with text editor support and such things
> wouldn't be the end of the world, but the value of that is pretty
> limited if you're explicitly opting them out of community maintenance.
>
> If the build files can be maintained in their own directory, and will
> only be updated by some set of people who are using them, why not track
> them in a separate source repo? This way when a breaking change happens
> your build system could still be made to work for the commits in between
> the time that happens and the GN files are updated, since you'd be
> maintaining a separate mapping of which versions work together.
>
> > - Having a few people care about the GN build means these people won't
> > improve the cmake build. I think this has some merit (but I did clean up
> > parts of the cmake build a bit while reading all our cmake code to create
> > the gn build for example, and I have several ideas about improving the
> > cmake build I want to implement at some point; and it's also not a given
> > that the folks who may or may not end up working on the GN build would
> have
> > worked on the cmake build). However, this is true independently of where
> > the GN build files are stored.
> >
> > - GN isn't a cmake replacement, for distro reasons and whatnot. I
> > wholeheartedly agree with this. Maybe GN will become better here, but
> cmake
> > is ubiquitous _today_, and it's backed by a company who's interested in
> > keeping it around, so it will be around for a long time. I think this is
> > cmake's biggest strength. That's why I'm not proposing on replacing the
> > cmake build. If the GN build at some point is way better (compiler-rt,
> > lldb, etc added; out-of-tree build support; GN itself gets a real distro
> > story; ...) then this _may_ be different in a few years, at which point
> we
> > could reconsider. I think this is unlikely to happen.
> >
> > (I still don't see that any of these problems are being solved by having
> > this in an overlay or a separate fork.)
> >
> > And again, It's a real possibility that we check this in and it turns out
> > the people who are interested in don't feel it's all that useful, it
> > bitrots, and we delete it again. That's cool, no harm done. I think we
> > should have a very high standard for replacing the supported build system
> > (cmake), but I think it's cool to have a low bar for people experimenting
> > with unsupported build systems, as long as they get deleted if not used.
> If
> > someone wanted to a, say, llvm/experimental/meson to see how that would
> > feel, I think that'd be super cool too for example.
> >
> > On Fri, Nov 2, 2018 at 2:06 AM Dean Michael Berris <
> dean.berris at gmail.com>
> > wrote:
> >
> >> I think this is how, after moving to a monorepo, it may be feasible to
> >> get this in a separate fork.
> >>
> >> Granted the Git Monorepo + GitHub happens, we can even make it so that
> >> the GN build is a branch on the official git repository, which can
> >> track the mainline development.
> >> On Fri, Nov 2, 2018 at 3:49 AM David Blaikie via llvm-dev
> >> <llvm-dev at lists.llvm.org> wrote:
> >> >
> >> > Any easy way to do this as some kind of overlay, so they GN files
> >> wouldn't live in the LLVM repository, but in a separate one?
> >> >
> >> > (this might avoid some of the community discussion - though would
> >> perhaps still likely have the issue I see as most significant: With a
> >> sufficient number of developers using GN, the rate of build breaks due
> to
> >> those developers missing CMake file updates might rise to be a bit of a
> >> drag on the LLVM project - though I don't think that's likely to ever
> be a
> >> huge deal, just an annoyance)
> >> >
> >> > On Wed, Oct 31, 2018 at 11:19 AM Nico Weber via cfe-dev <
> >> cfe-dev at lists.llvm.org> wrote:
> >> >>
> >> >> Hi,
> >> >>
> >> >> first things first: If you're happy with cmake, you can stop reading
> >> now. Nobody is proposing that LLVM moves off cmake, and nobody is
> proposing
> >> anything that's causing people using cmake more work.
> >> >>
> >> >> At the LLVM conference, I gave a lightning talk [1] about using GN
> [2]
> >> to build LLVM and clang. cmake is great for many use cases, but in my
> >> opinion local day-to-day development isn't one of them. So I wrote GN
> build
> >> files for LLVM and clang, enough to make `ninja check-llvm check-clang
> >> check-lld` build everything needed for these three test suites and that
> all
> >> tests pass. This works on Linux, Mac, Win hosts targeting X86, ARM,
> >> AArch64. You can see them at [3].
> >> >>
> >> >> I had been worried that it would be a lot of work to keep the build
> >> files up to date, but I've been using this for all my LLVM/clang/lld
> >> development the last 8 months, and it turned out to not be a big
> problem --
> >> LLVM's build files don't change very often, and GN build files are a
> >> pleasure to work with in my opinion.
> >> >>
> >> >> I gave the lightning talk just to talk about my personal workflow,
> but
> >> there was a lot of interest. We had a roundtable on the next day, and
> about
> >> 20 people said they'd be interested in using this for their development
> >> too. The main request was that the .gn files are checked in upstream, so
> >> that we can collaborate on keeping them working. Two to three orgs even
> >> said they'd be interested in moving their buildbots to GN.
> >> >>
> >> >> As mentioned at the top, the intention here is not to replace cmake,
> >> only to offer an opt-in alternative for people who are interested in it.
> >> Keeping the GN build going would be the responsibility of people using
> it,
> >> not of the general LLVM community. If this fails to find use and
> bitrots,
> >> we can easily remove it again.
> >> >>
> >> >> Are there any concerns with checking in GN files? I've put some
> initial
> >> docs for the GN build at https://reviews.llvm.org/D53944 ; it describes
> >> what the GN build is and is not, what its advantages are (speed and
> easier
> >> configurability), and some points about the philosophy behind the LLVM
> GN
> >> build.
> >> >>
> >> >> If folks are generally ok with GN files living in-tree, I'll start
> >> sending patches for gradually adding gn files through the regular review
> >> process.
> >> >>
> >> >> If having a BUILD.gn file in every directory being confusing is a
> >> concern, GN has the concept of a "secondary tree" so that all GN files
> >> could be below llvm/gn/tree/{llvm,clang,lld,...}.
> >> >>
> >> >> Cheers,
> >> >> Nico
> >> >>
> >> >> 1: https://llvm.org/devmtg/2018-10/talk-abstracts.html#lt2
> >> >> 2: https://gn.googlesource.com/gn , https://is.gd/gn_intro
> >> >> 3:
> >>
> https://github.com/llvm-project/llvm-project-20170507/compare/master...nico:gn
> >> , click "Files Changed" to see the GN files.
> >> >> _______________________________________________
> >> >> cfe-dev mailing list
> >> >> cfe-dev at lists.llvm.org
> >> >> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev
> >> >
> >> > _______________________________________________
> >> > LLVM Developers mailing list
> >> > llvm-dev at lists.llvm.org
> >> > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
> >>
> >>
> >>
> >> --
> >> Dean
> >>
> > _______________________________________________
> > LLVM Developers mailing list
> > llvm-dev at lists.llvm.org
> > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20181106/44088ddf/attachment.html>