develooper Front page | perl.perl5.porters | Postings from July 2010

Re: [PATCH] Configure probe for static inline

Thread Previous | Thread Next
Nicholas Clark
July 9, 2010 07:05
Re: [PATCH] Configure probe for static inline
Message ID:
On Fri, Jul 09, 2010 at 10:01:56AM -0400, Andy Dougherty wrote:
> On Fri, 9 Jul 2010, Nicholas Clark wrote:
> > On Thu, Jul 08, 2010 at 11:34:23AM -0400, Andy Dougherty wrote:
> > 
> > > Perl generally doesn't use gcc -ansi (except for dec_osf and irix) so 
> > > that's not a big issue, but given that we'll need to support __inline for 
> > > MSVC anyway, I will restructure the probe to make that easier to 
> > > incorporate, and might as well test for it anyway.
> > 
> > But it is possible to opt to turn -ansi -pedantic on (with -Dgccansipedantic
> > according to Porting/pumpkin.pod), which has been useful at trying to keep
> > the code portably clean when all you have is a hammer, er gcc.
> This would still work fine, it just would use 'static' instead of 'static 
> inline', though the compiler would be free to inline it anyway.  However, 
> it's academic anyway because I have revised the patch to try 
> __inline__ (and __inline and _inline) if "inline" doesn't work.

Nice. Thanks for this.

> > Last I checked, of the platforms I had access to, it was most useful on
> > FreeBSD, where the system headers are clean. (And cleaner than Solaris, which
> > was a shock. I don't currently have access to Solaris)
> Last time I investigated this (which was some time ago), the Solaris 
> headers didn't pass under -ansi -pedantic at least in part due to Sun's 
> efforts to ensure backwards compatibility.

Ah. That's a valid reason. Although they don't do as well as AIX.
Last time I had access to AIX (a couple of years ago), 5.000 would still build
out of the box on it. This isn't true on Solaris.

Nicholas Clark

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About