Date: Sun, 22 Feb 2015 17:32:23 +0100 (CET) From: Gerald Pfeifer <gerald@pfeifer.com> To: Thomas Zander <riggs@FreeBSD.org>, phd_kimberlite@yahoo.co.jp Cc: svn-ports-head@freebsd.org, svn-ports-all@freebsd.org, ports-committers@freebsd.org Subject: Re: svn commit: r379512 - in head/math/openblas: . files Message-ID: <alpine.LSU.2.11.1502221728200.4743@tuna.site> In-Reply-To: <201502211340.t1LDeKhn057393@svn.freebsd.org> References: <201502211340.t1LDeKhn057393@svn.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Saturday 2015-02-21 13:40, Thomas Zander wrote: > Log: > Add support for clang > > PR: 197385 > Submitted by: phd_kimberlite@yahoo.co.jp (maintainer) > -USE_GCC= yes So far, so good. > - -e 's+%%_GCC_RUNTIME%%+${_GCC_RUNTIME}+' \ > + -e 's+%%_GCC_VER%%+${_GCC_VER}+' \ This looks scary. Why do you replace one variable starting with an underscore by another variable starting with an underscore when in Unix-land this convention means "internal use; not to be used by regular users"? > Modified: head/math/openblas/files/patch-Makefile.system > ============================================================================== > -+FCOMMON_OPT += -Wl,-rpath=%%_GCC_RUNTIME%% > ++FCOMMON_OPT += -Wl,-rpath=%%LOCALBASE%%/lib/gcc%%_GCC_VER%% And here it becomes really worrisome. Instead of the internal variable that at least had the full path set in a transparent manner, now you construct this manually? Totally relying on implementation details that can (and likely _will_) change? And without USE_GCC set, _GCC_VER is not going to set either, is it? I'm afraid something is broken here. Can you please address it? Gerald
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?alpine.LSU.2.11.1502221728200.4743>