On 2/24/17 1:48 PM, Eric Fiselier wrote:
Inserting arbitrary feature macros into CMake should not be a
supported scenario because it results is macros, such as this one,
which are seemingly dead.
Good point.
Jon
/Eric
On Fri, Feb 24, 2017 at 1:44 PM, Jonathan Roelofs
<jonat...@codesourcery.com <mailto:jonat...@codesourcery.com>> wrote:
On 2/24/17 1:30 PM, Eric Fiselier via Phabricator wrote:
EricWF added a comment.
In https://reviews.llvm.org/D30339#685921
<https://reviews.llvm.org/D30339#685921>, @jroelofs wrote:
In https://reviews.llvm.org/D30339#685919
<https://reviews.llvm.org/D30339#685919>, @rmaprath wrote:
Perhaps change `config.h` and remove the definition
there and adjust other places accordingly?
The current form is very easy to trip over.
Eric's point is that LIBCXXABI_BAREMETAL is a 0/1 flag,
not a defined/not-defined flag. Please don't change from
one form to the other... it's disruptive to build systems.
I actually think it's better to maintain consistency between
libc++ and libc++abi. And libc++ never uses 0/1 flags. So I
would rather see a fix in `config.h`.
Frankly I don't care that it is disruptive to build systems
unless it's the build system owned by LLVM.
What I really care about is the interface between the build system
owned by LLVM, and the one driving it.
Jon
Repository:
rL LLVM
https://reviews.llvm.org/D30339 <https://reviews.llvm.org/D30339>
--
Jon Roelofs
jonat...@codesourcery.com <mailto:jonat...@codesourcery.com>
CodeSourcery / Mentor Embedded
--
Jon Roelofs
jonat...@codesourcery.com
CodeSourcery / Mentor Embedded
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
