Date: Tue, 19 Jun 2012 05:08:55 -0400 From: "Thomas Mueller" <mueller230@insightbb.com> To: freebsd-questions@freebsd.org Cc: David Naylor <naylor.b.david@gmail.com>, Volodymyr Kostyrko <c.kworr@gmail.com>, Matthew Seaman <matthew@FreeBSD.org> Subject: Re: Why Clang? Message-ID: <14.FA.06229.7A140EF4@smtp01.insight.synacor.com>
next in thread | raw e-mail | index | archive | help
from David Naylor: > I am the one who sends these persistent messages. Some users of my packages > reported that wine didn't run due to a clang compiled world. I never verified > them (although I got multiple reports). With the updates to clang it may have > also been corrected. > I attributed the problem to clang miscompiling a library in base used by wine > and Volodymyr, I think, confirms this: I only have other people's experience on this issue, need to test this, but want to keep a GCC-compiled world for now, at least for a production system. This would not stop me from trying Clang on an experimental/testing installation, such as HEAD, where the basic intent is development. >From Volodymyr Kostyrko: > Thomas Mueller wrote: > >Now one concern is wine not working when Clang is used to "make buildworld". > For me I'm just waiting on toolchain stabilization as both this one and > (open|libre)office fail because of libgcc_s compiled with clang on amd64. I guess that's why I want to keep at least one GCC-compiled world for now. Like it or not, Linux is by far the leading open-source OS, and most of the ports are originally developed with mainly Linux in mind. Linux software development is GCC-centric, I don't know if there is any work with Clang in Linux. Now how will I know whether GCC or Clang is the default compiler for building the world and kernel, and for ports? Not that I want to avoid Clang, just don't want to be caught by surprise. Tom
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?14.FA.06229.7A140EF4>