Date: Sun, 24 Apr 2016 13:24:22 -0400 From: Steve Wills <swills@FreeBSD.org> To: Mark Millard <markmi@dsl-only.net> Cc: freebsd-ports@freebsd.org, FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: port's svn commit: r413746 - in head "many ports: mark broken on powerpc64": for what toolchains? Message-ID: <571D0146.5060200@FreeBSD.org> In-Reply-To: <F7E6ED93-A73D-406D-A7BF-B1B80C61871F@dsl-only.net> References: <34C0599F-044B-46ED-AF60-0F0E98876E2F@dsl-only.net> <571C0297.3050801@FreeBSD.org> <28FDFFB4-02CC-40CB-ACAC-828BA8E71A37@dsl-only.net> <00621189-D577-4E3F-8BAB-4B315B690209@dsl-only.net> <571CC2F2.2060601@FreeBSD.org> <F7E6ED93-A73D-406D-A7BF-B1B80C61871F@dsl-only.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi,
On 04/24/16 10:16 AM, Mark Millard wrote:
>
> For all the port update activity (including ruby) I used gcc49, /etc/make.conf being:
>
> # more /etc/make.conf DEFAULT_VERSIONS+=perl5=5.22
> WRKDIRPREFIX=/usr/obj/portswork
> WITH_DEBUG=
> WITH_DEBUG_FILES=
> MALLOC_PRODUCTION=
> #
> #
> # For trying gcc49...
> #
> CC=/usr/local/bin/gcc49
> CXX=/usr/local/bin/g++49
> CPP=/usr/local/bin/cpp49
> . . . (binutils macros omitted here) . . .
>
>
> (I do not know if lang/gcc [or lang/gcc48] would work or not. I
> prefer a tool chain with a more modern C++ available but gcc49 is not
> yet what lang/gcc builds.)
>
>
>
> I've seen notation like:
>
> USE_GCC= 4.9+
>
> in port Makefiles. Also notation like:
>
> .if ${ARCH} == powerpc64
>
> and:
>
> .if ${ARCH} == "powerpc" || ${ARCH} == "powerpc64"
>
>
> So may be the extra notation in the Makefile(s) in question could be something like:
>
> # clang 3.8.0 and before is still broken in various ways for powerpc and powerpc64:
> .if ${ARCH} == "powerpc" || ${ARCH} == "powerpc64"
> USE_GCC= 4.9+
> .endif
>
Yep, this sounds right to me. I will test this with at least lang/ruby22
and lang/gcc6-devel when my current build finishes, or sooner if I get
impatient. :)
> I list both powerpc variants because powerpc and powerpc64 both have
> clang problems making buildworld a no-go by default and if gcc 4.2.1
> rejects a port for one it would normally also reject for the other.
> There may be other ${ARCH} values that would also be appropriate
> because they are also stuck at gcc 4.2.1 .
Makes sense.
> I do not claim to know necessary vs. sufficient status: more might be
> needed for some configurations (rpath issues? mixture of libraries
> compiled by distinct gcc's?). But I expect that the above should be
> better than being marked broken.
We'll find this out when we test! :)
Thanks,
Steve
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?571D0146.5060200>