[webkit-dev] Build slave for JSCOnly Linux MIPS

Tue Apr 19 11:33:04 PDT 2016

19.04.2016, 21:15, "Filip Pizlo" <fpizlo at apple.com>:
> I did a quick look over the trac query of GCC 4.8 changes that you provided. None of the ones I looked at were scary but they were annoying. They seemed to be things like:
>
> - Sometimes saying { } to initialize a variable doesn’t work.
> - Sometimes you need to say “const”.
> - Sometimes you need to play with variables to get around internal compiler errors.
>
> I didn’t find any cases of GCC 4.8 not supporting a language feature that we want to use. Do you think that’s correct?

According to [1], GCC provides complete C++11 feature list since 4.8.1. However, it fails to compile FTLLazySlowPathCall.h, see complete set of diagnostics in [2].

There is another minor bug: 4.8 does not allow aggregate initialization for structs which have deleted constructors [3].

[1] https://gcc.gnu.org/projects/cxx-status.html#cxx11
[2] http://pastebin.com/ikyDTZ9s
[3] https://bugs.webkit.org/show_bug.cgi?id=155698
    https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52707

>
> -Filip
>
>>  On Apr 19, 2016, at 11:02 AM, Michael Catanzaro <mcatanzaro at igalia.com> wrote:
>>
>>  Hi,
>>
>>  On Mon, 2016-04-18 at 17:27 -0700, Filip Pizlo wrote:
>>>  I am sympathetic to the principle that we should support the
>>>  compilers that ship on the most popular versions of Linux.
>>
>>  Great. :)
>>
>>>  I’d like to understand if that argument is sufficient to support GCC
>>>  4.8.
>>>
>>>  Can you clarify, is it the case that if I installed the latest stable
>>>  Fedora, I’d get GCC 4.8?
>>
>>  No, all currently-supported versions of Fedora include GCC 5 (only).
>>  Different distros have very different release cycles and policies for
>>  compiler upgrades. Fedora releases roughly every six months, and each
>>  release is supported for roughly 13 months. GCC releases once per year.
>>  The GCC developers coordinate with Fedora release planning to time GCC
>>  releases to coincide with spring Fedora releases; in the winter before
>>  a new GCC release, we rebuilt all of Fedora with the GCC beta so the
>>  GCC developers can collect bug reports. So we will never have issues
>>  with Fedora, as the oldest Fedora will be at most one year behind
>>  upstream GCC. (Note that I co-maintain the WebKitGTK+ package there and
>>  I'm making sure all supported Fedoras get updates.)
>>
>>  But Fedora is exceptional in this regard. Other distros are supported
>>  for much longer than 13 months (5 years for Ubuntu LTS and newly also
>>  for Debian, 10 years for enterprise distros) and therefore have much
>>  older compilers. The question is where do we draw the line. We
>>  obviously cannot support a 10 year old distro; those are maintained by
>>  rich corporations, and if they cared about WebKit security, they could
>>  take responsibility for that. We could handle 5 years, but do we really
>>  want to? (It's clear Apple doesn't.) It's really inconvenient to not
>>  have access to newer dependencies or language features for so long. We
>>  might start by saying that we only support the latest release of [list
>>  of major distros that have recently been shipping WebKit updates]. Most
>>  of these distros are currently built using GCC 4.9, though they might
>>  have GCC 5 or GCC 6 packaged as well, but not used by default. The big
>>  one still using GCC 4.8 is openSUSE.
>>
>>  We don't *need* to consider Ubuntu right now, because they rarely ever
>>  take our updates, nor Debian, because they never take our updates. I
>>  think WebKit updates for Debian is all but totally a lost cause, but
>>  I'm kinda still hopeful for Ubuntu, so I'd like to keep them in mind.
>>
>>  Also, different distros have different policies on using alternative
>>  compilers. E.g. in Fedora we are usually required to always use
>>  Fedora's GCC, and only one version is available at a time... but if a
>>  package *really* has no chance of being built with GCC, we're allowed
>>  to use Fedora's Clang instead. I'm not sure what the policies are for
>>  Debian and Ubuntu, but they always have available a newer GCC than is
>>  used for building packages, and until recently were using Clang to
>>  build Chromium, so alternative compilers must be permitted at least in
>>  exceptional cases. I was trying to convince the openSUSE folks to use
>>  Clang to build WebKit, to avoid the GCC 4.8 issue, but they were not
>>  enthusiastic. (But consider that all these distros will have older
>>  versions of Clang as well.)
>>
>>  Now, whether openSUSE is important enough on its own to justify holding
>>  back or lowering our GCC requirement... maybe not. But anyway, since we
>>  have significant contributors like Konstantin stuck with GCC 4.8, and
>>  since this doesn't require giving up on any significant language
>>  features, I think it's OK. If it's only a little work to support that
>>  compiler (on the level we already have in trunk), I think it's a good
>>  idea.
>>
>>  But there is another problem here. openSUSE seems to have no intention
>>  of upgrading to a newer GCC anytime soon, because they have started to
>>  inherit core packages like GCC from the SUSE enterprise distro. So I
>>  might need to negotiate with them if it would be possible to build
>>  WebKit with clang after all.
>>
>>>  Can you clarify what you mean by “backport”? I’m trying to get a
>>>  picture of how your releases work. For example, are you saying that
>>>  RHEL wouldn’t take a security update that you backported, or that
>>>  they won’t invest energy into backporting it themselves?
>>
>>  We don't try to convince distros to take individual security fixes as
>>  patches, because there are way too many for that to be practical. We
>>  want them to take our tarball updates.
>>
>>  In that mail I was saying that RHEL won't invest energy into
>>  backporting things themselves downstream; consider that we have about
>>  100 security fixes per year, backporting from trunk needs to be handled
>>  upstream so this can be shared among distros, rather than separately by
>>  each distro that wants to provide WebKit updates. Our upstream
>>  WebKitGTK+ releases work like this: every February and August, we
>>  branch off of trunk; this forms a new stable branch, which gets
>>  released in March/September. We then cherry-pick fixes to that branch
>>  and make releases off of it for the next seven months or so. Our goal
>>  is to convince distros to take these releases, because it's the only
>>  practical way for them to get security updates. I've recently had some
>>  mixed success with this; a couple big names like Mageia and openSUSE
>>  recently started taking our updates.
>>
>>  Some distros like Debian refuse to take any version upgrades at all,
>>  and want to fix everything with downstream patches. Since that is not
>>  practical for WebKit, they have adopted a policy of no security support
>>  for WebKit. Ubuntu leans towards this as well, but occasionally they do
>>  take our updates; I'm hoping that might become more common.
>>
>>  (RHEL is a bit of a special case in that its old enough that all apps
>>  in RHEL are using WebKit1, which we don't support anymore, so there's
>>  no value in taking our updates.)
>>
>>>  How many changes are required to make GCC 4.8 work? I think this
>>>  will provide important context for this discussion.
>>
>>  I guess it's working already and we only need to remove the build error
>>  when it's detected, because Konstantin has been committing GCC 4.8
>>  fixes throughout the tree:
>>
>>  http://trac.webkit.org/search?q=4.8&noquickjump=1&changeset=on
>>
>>  Anyway, I do not strongly request that we drop the GCC requirement to
>>  GCC 4.8, though I think that would be fine; just please, we should keep
>>  these issues in mind when upgrading our compiler requirement in the
>>  future.
>>
>>  Michael
>
> _______________________________________________
> webkit-dev mailing list
> webkit-dev at lists.webkit.org
> https://lists.webkit.org/mailman/listinfo/webkit-dev

-- 
Regards,
Konstantin