Cutting down the number of platform checks

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Cutting down the number of platform checks

Reid Kleckner-2
On Tue, Jul 23, 2013 at 5:53 PM, Óscar Fuentes <[hidden email]> wrote:
Yes. What you are seeing are the platform checks, where the build system
looks for the presence of functions, headers, etc and then generates a
configuration file with that information.

I've been meaning to cut down on the number of these because they are super slow and wasteful.  Some of them are dead and can be removed without discussion.

Some of them are used inconsistently, like HAVE_STRING_H.  Do we really support any platform that lacks a <string.h>?  All of Errno.cpp is in an ifdef for this macro, but I suspect we include string.h elsewhere unconditionally.

Is everyone OK with eliminating checks for headers and symbols that we use unconditionally anyway?  (assert.h, mempcy, etc)

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
Reply | Threaded
Open this post in threaded view
|

Re: Cutting down the number of platform checks

Eric Christopher
Sure. Preapproved if you feel the need for autoconf. Let me know if
you need/want help regenerating.

-eric

On Tue, Jul 23, 2013 at 3:30 PM, Reid Kleckner <[hidden email]> wrote:

> On Tue, Jul 23, 2013 at 5:53 PM, Óscar Fuentes <[hidden email]> wrote:
>>
>> Yes. What you are seeing are the platform checks, where the build system
>> looks for the presence of functions, headers, etc and then generates a
>> configuration file with that information.
>
>
> I've been meaning to cut down on the number of these because they are super
> slow and wasteful.  Some of them are dead and can be removed without
> discussion.
>
> Some of them are used inconsistently, like HAVE_STRING_H.  Do we really
> support any platform that lacks a <string.h>?  All of Errno.cpp is in an
> ifdef for this macro, but I suspect we include string.h elsewhere
> unconditionally.
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Support/Errno.cpp?revision=167191&view=markup
>
> Is everyone OK with eliminating checks for headers and symbols that we use
> unconditionally anyway?  (assert.h, mempcy, etc)
>
> _______________________________________________
> LLVM Developers mailing list
> [hidden email]         http://llvm.cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>

_______________________________________________
LLVM Developers mailing list
[hidden email]         http://llvm.cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev