On 12/12/2012 09:10 AM, Viktor Mihajlovski wrote:
On 12/07/2012 06:57 PM, Eric Blake wrote:
>
>
> No. Don't use index(), as it is not guaranteed to exist.
>
> What we should _really_ do is provide a change in
> m4/virt-compile-warnings.m4 that disables -Wlogical-op if compiling with
> a broken gcc, now that newer gcc doesn't force its stupidity on a
> sensible strchr() usage. I'll attempt that patch independently, and
> then your patch can go on top using strchr() from the get-go.
>
On a second thought I didn't like to disable the warning
for all sources as I still would like to be informed about
potential issues.
Don't worry about it. There are enough people compiling with newer gcc
where the warning is not disabled that you won't be introducing in problems.
I've send out a new version of the patch (in fact a new series)
disabling the warning with a pragma only in the affected file.
But isn't the problem that the use of pragmas to silence broken warnings
is specific to newer gcc, while you are trying to silence the warning on
older gcc? I guess I'll have to look at what you did; and I still need
to work up my patch for the warning avoidance at configure time anyway.
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org