Date: Sun, 15 Jun 2014 09:04:24 -0700 From: Justin Hibbits <chmeeedalf@gmail.com> To: Nathan Whitehorn <nwhitehorn@freebsd.org> Cc: FreeBSD Mailing List <freebsd-ports@freebsd.org>, FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: PowerPC Packages Message-ID: <20140615090424.56f1caa7@zhabar.att.net> In-Reply-To: <539DC0C5.60603@freebsd.org> References: <539DC0C5.60603@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 15 Jun 2014 08:50:29 -0700 Nathan Whitehorn <nwhitehorn@freebsd.org> wrote: > Over the last few weeks (2 cores makes the build take a long time!), > I've built a complete set of 32-bit PowerPC packages for 10-STABLE, > which can be found at > http://people.freebsd.org/~nwhitehorn/powerpc-10-packages. This > address can be used as a pkg repository, but also contains a > post-mortem of what fails to build and why if you look at it in a web > browser, which may be of interest to porters. Note that these > packages are unofficial and so are accompanied by no guarantees > whatsoever from anyone. The packages are still slowly uploading, but > the logs and debug information are there already. > > Over all, 19443 ports succeeded. 1116 failed for some reason (this > number includes things marked BROKEN) and 3867 didn't start because > ports they depend upon failed. The largest issues are Boost and QT4 > webkit. Boost has failed because it seems to unconditionally prefer > clang to gcc when both are present. On PowerPC, both are installed > but GCC remains the default due to some remaining issues with clang. > Boost ignores the default, runs into the issues, and fails. Maybe it > should be forced to use whatever "cc" is? Several other ports run > into this issue. QT4 webkit seems to have caused GCC to segfault for > some reason. > > The website has a list of failed builds sorted by the number of ports > they are blocking, both directly (ports with this ports in their > *_DEPENDS line) and indirectly (all of those, plus all the ports that > depend on this port's children). Also listed is the point at which > the job failed and a link to the build log. "JOB_FAILURE" means that > the build or packaging step failed. "PRE_SCRIPT_FAILURE" means that > the job failed in the "make fetch" stage, which includes ports marked > BROKEN. > > I am more than happy to try out patches if porters without access to > PowerPC systems want to fix any of these. > -Nathan This is awesome! To fix the boost build, bapt suggested to me on IRC a while back to add 'FAVORITE_COMPILER=gcc' to my make.conf, so that it always prefers gcc. I've been working on ppc32 PIC for the last week or so when I have time, and hope to have it done in the next couple weeks at which point this hack should be moot for us. - Justin
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20140615090424.56f1caa7>