[cfe-dev] What are the plans for creating a Windows Binary Release of Clang?
Richard Smith
richard at metafoo.co.uk
Mon Nov 11 13:53:21 PST 2013
On Mon, Nov 11, 2013 at 12:29 PM, Alp Toker <alp at nuanti.com> wrote:
>
> On 11/11/2013 19:59, Richard Smith wrote:
> > On Mon, Nov 11, 2013 at 10:09 AM, Alp Toker <alp at nuanti.com
> > <mailto:alp at nuanti.com>> wrote:
> >
> >
> > On 11/11/2013 17:57, Chandler Carruth wrote:
> > >
> > > On Mon, Nov 11, 2013 at 7:01 AM, Alp Toker <alp at nuanti.com
> > <mailto:alp at nuanti.com>
> > > <mailto:alp at nuanti.com <mailto:alp at nuanti.com>>> wrote:
> > >
> > > Hello Devin,
> > >
> > > We're producing a Windows version of clang at http://clang.org
> >
> >
> > Who is this "we"?
> >
> > We (the actual LLVM project) are producing a Windows version of clang
> > at http://llvm.org/builds
>
> I did qualify that this download is unofficial. The site is a place to
> upload clang-related patches and builds for testing.
>
> There are different groups of "we" posting all the time on this list,
> whether it's Apple, Google, FreeBSD or Debian and I can only speak for
> myself and Nuanti.
There's also the "we" representing the community consensus that forms after
discussion and review on these lists.
> > I think it's really unfortunate that you're running your own website
> > > and rolling your own builds. =/ So others on the list know, this is
> > > not in any way part of the LLVM open source project, it is not
> being
> > > maintained by the community at large but by some individuals (Alp
> at
> > > least). Putting it at 'clang.org <http://clang.org>
> > <http://clang.org>' is a best misleading.
> >
> > And llvm-reviews.chandlerc.com <http://llvm-reviews.chandlerc.com>
> > is maintained by the community, *ahem*?
> >
> >
> > Actually, yes, it is. And it was discussed with the community. And the
> > domain name does not suggest that it is somehow official (although it
> > actually is much more official and community-reviewed than clang.org
> > <http://clang.org>). And it doesn't provide Windows binaries to
> > download that people might think are somehow official.
> >
> > I really have no idea why you did this without discussing it with the
> > Clang and LLVM community first. While I'm sure you had the best of
> > intentions, this really isn't a good way of interacting with the rest
> > of us, and if the status quo is maintained, this will generate
> > hostility towards you.
>
> Nothing's been done. There's been something on the site for probably a
> year?
I was talking about putting up the site in the first place, not whatever
tweaks you might have made recently. It is (of course) completely
reasonable for you to have a website that talks about clang, and links to
our website, and so on -- the point of contention is *only* that the
website might be confused with an official part of the project, both due to
its domain name and because it ends by saying "part of the LLVM Open Source
project".
> Here's what I would suggest as some plans for moving forward:
> >
> > Option 1:
> > You explain what you're trying to achieve with this website, what's
> > missing from the existing website, and how we can address that.
>
> Here's the background: We were chatting on the #LLVM IRC channel during
> the devmeeting, and put up some balloons on that page to celebrate the
> commit you did with Faisal completing C++14.
>
> It was a significant moment for a few of us who made contribution's to
> clang's C++0x support, I think possibly before you were around on the
> project?
>
If this is supposed to provide generally-useful information about clang,
why is *another* website necessary, in addition to the official one? Is our
existing site (for instance) too hard to navigate? If this site is solely
intended to be your own personal site about clang, it should make that
clear, and I'm grateful that you're willing to do that.
[The only novel content on the site seems to be the additional Windows
binaries, containing some revision of clang ToT plus unspecified additional
changes. I find it really worrying to have such binaries on a website that
an unsuspecting person might think is operated by the Clang project. What
if such a binary (accidentally or maliciously) contained a virus, for
instance?]
As for the download, a couple of days there was a need to upload test
> builds with the patches from the PR, and the site was already there.
>
>
> > Someone (ideally you, since you seem interested in driving this)
> > provides patches to our existing website to achieve these goals.
>
> A patch adding balloons? I'm not sure what you see on clang.org that's
> significant -- it's a homepage for clang-related odds and ends.
> You forward clang.org <http://clang.org> to clang.llvm.org
> > <http://clang.llvm.org> (or some subpage of it).
> >
> > Option 2:
> > You add some text to clang.org <http://clang.org> explaining who is
> > running it, that it is not official and not sanctioned by the upstream
> > LLVM and Clang projects, and why it exists.
>
> Sure, let's do this. Though I don't really feel like celebrating r194188
> after this discussion :-/
>
On the other hand It's been encouraging to hear from people who've been
> enjoyed the builds in the meantime and reporting back. The real solution
> is to actually get these fixes landed.
>
I entirely agree, and I think this is a solid and productive way to make
progress. Can you point out which fixes they are, so we can work together
to get them in fit shape to be pushed into svn?
> Cheers,
> Alp.
>
>
> >
> > But yes, let's add some links from the documentation on llvm.org
> > <http://llvm.org> if you
> > think it'll clear things up. It's just there's not much there :-)
> >
> >
> > >
> > >
> > >
> > > The main difference from the installer on LLVM.org is that
> > this one is
> > > dedicated to the clang frontend and aims to test and
> > stabilise the
> > > latest patches, and is code signed for use in a corporate
> > > environment etc.
> > >
> > >
> > > And that it isn't being produced or maintained by the LLVM open
> > source
> > > project. =[ I think the authors should make that much more
> > clear, and
> > > I hope that instead of running their own website they instead help
> > > contribute to the open source build, test, and release process.
> >
> > All these patches are on the bug tracker, mailing list or already
> > committed. There are a few issues that are showing up as regular
> dupes
> > on the bug tracker and need work including PR17403 and these
> > builds help
> > test whether features are good to land for 3.4.
> >
> > Since yesterday the builds are also helping answer questions like
> > whether Visual Studio 2013 produces builds compatible with Windows XP
> > (answer is No right now -- we need to find the right flags and see
> how
> > much C++11 we can get out of it).
> >
> > > (Note, I know that Alp at least has contributed a few patches, but
> > > certainly hasn't been responsible for the overwhelming majority
> > of the
> > > work to bring Clang to Windows.)
> >
> > Nobody ever claimed that -- it's absolutely a group effort. My work
> > since 2010 on the LLVM project has generally been in the C++ frontend
> > targeting Unix so this is new to me as well. You could have looked
> > that
> > up easily.
> >
> > That said we did go to some expense to start setting up this
> > builder and
> > I'm not too interested in unconstructive responses at this point.
> >
> > Alp.
> >
> >
> > --
> > http://www.nuanti.com
> > the browser experts
> >
> > _______________________________________________
> > cfe-dev mailing list
> > cfe-dev at cs.uiuc.edu <mailto:cfe-dev at cs.uiuc.edu>
> > http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
> >
> >
>
> --
> http://www.nuanti.com
> the browser experts
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20131111/939b1c80/attachment.html>
More information about the cfe-dev
mailing list