|
|
Boost Testing : |
Subject: Re: [Boost-testing] minion-clang runner: incorrect version information
From: Christopher Jefferson (chris_at_[hidden])
Date: 2011-02-13 05:31:16
On 13 Feb 2011, at 06:58, Bryce Lelbach (wash) wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Who is the minion-clang runner? The version specified for clang on the
> regression page is incorrect; there is no clang version "4.2.1". That is
> probably a reference to the version of the GNU stdlib/system version of GCC.
> This is erronous and potentially confusing. I would like to know what version
> of clang/llvm each of your test runners is supplied with, and I would
> appreciate it if you would change the compiler name that is supplied to
> regression.py.
Just to say, I run this tester, for future reference.
It currently runs the day's svn head, which does lead to occasional breakage when something goes badly wrong with clang. However the last released version of clang isn't really good enough to run all of boost.
Hopefully when the next version is released, I will track that instead.
I am currently extending the list of clangs I test to (I will document this in comment.html) include clang in c++0x mode (with the gnu++0x flag, slightly confusingly) and with the libc++ standard library. These currently fail quite a lot, apologises for any confusion they might cause.
Chris