correctly detect presence of __sync_fetch_and_add

[unicolet]

It seems that gcc will provide a stub function on architectures that don't support atomic operations like __sync_fetch_and_add unless the argument is not an int32 and its return value is not used. This breaks mapserver compilation because prior to this pr the detection code did not check for the same exact conditions under which __sync_fetch_and_add is used in mapserver internals.

This pr improves the detection code so that situations like those described above can be handled correctly.

[tbonfort]

@unicolet is this good for 6.2.0 ?

[unicolet]

I think so, I was waiting for confirmation by the users on the ML

On Fri, Oct 19, 2012 at 3:07 PM, Thomas Bonfort notifications@github.comwrote:

@unicolet https://github.com/unicolet is this good for 6.2.0 ?

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/4502#issuecomment-9600263.

[AndreHeinecke]

I found this patch through google after getting the build error fixed by this.

I successful used (and needed) this patch to compile mapserver 6.2.0-rc1on an x86 centos 5 system.
Thanks.

[unicolet]

Andre, thanks for your feedback.
We will merge the pull request asap.

Regards,
Umberto

On Thursday, October 25, 2012, Andre Heinecke wrote:

I found this patch through google after getting the build error fixed by
this.

I successful used (and needed) this patch to compile mapserver 6.2.0-rc1on
an x86 centos 5 system.
Thanks.

—
Reply to this email directly or view it on GitHubhttps://github.com//pull/4502#issuecomment-9783169.