Are ports supposed to build and run on 10-CURRENT?
decke at FreeBSD.org
Fri Jun 14 15:49:51 UTC 2013
On Fri, Jun 14, 2013 at 2:36 PM, Michael Gmelin <freebsd at grem.de> wrote:
> On Thu, 13 Jun 2013 08:24:18 +0200
> Matthias Apitz <guru at unixarea.de> wrote:
>> El día Thursday, June 13, 2013 a las 08:07:14AM +0200, Eitan Adler
>> > On Thu, Jun 13, 2013 at 3:15 AM, Michael Gmelin <freebsd at grem.de>
>> > wrote:
>> > > So my question is: Are we port maintainers now really supposed to
>> > > make ports work with CURRENT?
>> > This is generally up to the maintainer; however many committers run
>> > -CURRENT and test on that by default.
>> This was a widely general question and a general answer too;
>> I'm afraid that not all committers run tests with -CURRENT and on both
>> architectures (amd64 and i386);
>> at least for KDE4 I can confirm that it is unable to build on -CURRENT
>> r250588 i386 due to:
>> 1. certain required ports does not compile with clang
>> 2. for kdelibs4 (and others) the tool automoc4 SIGSEGV randomly (i.e.
>> for different source files and if you run it again it works)
>> all this on a SVN clean ports tree;
>> the current situation with CURRENT/clang/ports is highly a concern;
> To be perfectly honest, the fact that some committers seem to *solely*
> test on CURRENT seems like bad QA practice to me.
This is true and also the reason why ports on CURRENT were always not
supported but only on best effort basis. This is not only the responsibility
of ports people but ALSO each and every src committer. We usually just
have to catch up things that change on current and break a number of ports.
Things like changing the default compiler, changing version numbers to
something that break each and every autoconf dependent port are just
the well known problems. Many ports break because of smaller changes
in the toolchain and libraries and it takes some time to fix all affected ports.
If that is something you don't want then CURRENT is not suitable for you.
More information about the freebsd-ports