Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 30 Dec 2019 13:52:26 +1000 (+10)
From:      Gerald Pfeifer <gerald@pfeifer.com>
To:        Tobias Kortkamp <tobik@freebsd.org>
Cc:        ports-committers@freebsd.org, svn-ports-all@freebsd.org,  svn-ports-head@freebsd.org
Subject:   Re: svn commit: r517725 - head/lang/gcc9
Message-ID:  <alpine.LSU.2.21.1912301352200.13725@anthias.pfeifer.com>
In-Reply-To: <20191222055804.GA251@urd.tobik.me>
References:  <201911160812.xAG8CZEi075212@repo.freebsd.org> <20191214155544.GA32673@urd.tobik.me> <alpine.LSU.2.21.1912221155590.3227@anthias> <20191222055804.GA251@urd.tobik.me>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, 22 Dec 2019, Tobias Kortkamp wrote:
>>> when can we enable PLUGINS by default?
> It unlocks the ability to package GCC plugins (like the one in
> security/afl++) without custom builds of GCC.  I would like to
> enable afl++'s GCC option by default, but a prerequisite of that
> is that PLUGINS is turned on by default in lang/gcc9.
> 
> Are there any downsides in enabling PLUGINS by default?

One downside is that the GCC plugin APIs explicitly are not stable, 
so can (and will) change with new versions of GCC.

Hence other ports using them will either need to pin to a fixed 
version of GCC (USE_GCC=9) and maybe become maintenance challenges
that way or become a blocker when we are next looking into updating 
the default version of GCC in Mk/bsd.default-versions.mk.

It's this kind of dependency that has made the later rather painful 
for the past years and iterations, though I do happily acknowledge
that you have been very supportive and helpful there.  Plus we have 
managed to fully catch up, for the first time in a long while. :-)

So I have gone ahead and enabled plugins for lang/gcc10-devel and
lang/gcc9-devel with recent snapshots and plan on letting lang/gcc9
follow at one point in January.


Why a bit later?  In my experience maintaining the lang/gcc* ports
"you never know". ;-)  Sometimes it's non-x86 architectures (notably
powerpc has kept us somewhat busy this year), sometimes it's -CURRENT, 
sometimes simply broader usage, and letting things settle a few weeks 
via gcc9-devel has been proven useful.

Makes sense?

Gerald



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.LSU.2.21.1912301352200.13725>