[PATCH] Don't set $LIB if we're not targetting windows.

Filipe Cabecinhas filcab+llvm.phabricator at gmail.com
Thu Feb 19 12:11:20 PST 2015


I may be misunderstanding, but:
It seems to me that, most likely, you don't want INCLUDE, since you want to
be sure of all the headers you include (and tests depend on a minimum set
of headers).

But you want LIB (on Windows targetting Windows) since that's where you can
find kernel32.dll and others.

LLVM tests only run on the host, so we don't have the “problem” of Windows
targetting some other platform. Because of that, we can simply forward LIB.

 Filipe

On Thu, Feb 19, 2015 at 9:52 AM, Hans Wennborg <hans at chromium.org> wrote:

> I'm not sure I understand what's being fixed here.
>
> LLVM's test/lit.cfg seems to just set LIB unconditionally
> (test/lit.cfg:75).
>
> A little higher up in this file we're conditionally putting INCLUDE in
> possibly_dangerous_env_vars, which seems to suggest that it that one also
> gets propagated through (but I can't see where that happens). Shouldn't we
> have the same logic for INCLUDE and LIB?
>
>
> http://reviews.llvm.org/D7739
>
> EMAIL PREFERENCES
>   http://reviews.llvm.org/settings/panel/emailpreferences/
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20150219/93231c5d/attachment.html>


More information about the llvm-commits mailing list