[cfe-commits] r67418 - in /cfe/trunk: include/clang/Driver/HostInfo.h lib/Driver/Driver.cpp lib/Driver/HostInfo.cpp test/Driver/phases.c tools/ccc/ccclib/Driver.py tools/ccc/ccclib/HostInfo.py

[Daniel Dunbar]

On Sat, Mar 21, 2009 at 10:48 AM, Chris Lattner <clattner at apple.com> wrote:

> On Mar 21, 2009, at 8:47 AM, Daniel Dunbar wrote:
>
> I don't understand this comment. On Darwin '<driver> foo.s' should run the
> preprocessor, then assemble. It doesn't really have anything to do with -E
> (although, due to this behavior, -E on a .s file on Darwin runs the
> preprocessor, and elsewhere will warn that the input file is unused).
>
>
> Wow, that is really weird and surprising.  Do you know if this is
> documented anywhere?
>

Indeed. And unfortunate!

It's not entirely weird, though. This matches the behavior with .S, and I
presume the historical reason Apple did this was coming from a case
insensitive file system. I don't believe it is document anywhere, although
the Apple as man page mentions it indirectly.

See gcc.c, and search for "This is kind of lame". :)

-Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20090321/61119ed0/attachment.html>