[Gluster-devel] Gluster and GCC 5.1

[Niels de Vos]

On Mon, Jun 29, 2015 at 03:00:26PM -0400, Kaleb S. KEITHLEY wrote:
> On 06/29/2015 02:45 PM, Kaleb S. KEITHLEY wrote:
> >>
> >>> E.g. by building on Fedora Rawhide last year, reporting the bugs,
> >>> and getting them fixed before Rawhide turned into Fedora 22.
> >>
> >>
> >> Immediately after the Fedora 22 release, we had a similar issue
> >> reported from a user with gcc v5 and was fixed through [1].
> >>
> > 
> > Yes, there were still some, and/or some had crept (back) in, when
> > Fedora22 was released.
> > 
> > I think we're still finding some more. Not all of the developers are
> > using Fedora 22 yet. I know a few that are still on Fedora 19!
> > 
> > Niels and I had a brief chat a while back about adding a Fedora 22 VM to
> > Jenkins to catch these. I don't think anything ever happened though.
> 
> Or perhaps we could just get everyone to stop using 'inline'

YES, THIS! I think there are only very few exceptions where we should
use suggest to the compiler what to do. Most compilers are smart enough
to make the right decisions.

It also prevents weird corner cases, like what happens to alloca() when
a function is inline'd? If you are not sure about when you could use
inline, you should rather not use it ;-)

Niels