Re: Allow to work on systems without Fcntl::flock implementation.

[Ralf Wildenhues]

* Eric Blake wrote on Wed, Sep 09, 2009 at 03:08:18AM CEST:
> According to Ralf Wildenhues on 9/8/2009 12:44 PM:
> > * Paolo Bonzini wrote on Tue, Sep 08, 2009 at 01:50:39PM CEST:
> >> On 09/08/2009 07:05 AM, Ralf Wildenhues wrote:
> >>> The patch is a bit of a hack, as it would be cleaner to put this right
> >>> into XFile.pm.  However, that would require us to either fork from the
> >>> Automake copy of the file, or also integrate this there.  WDYT?
> >> Why not? (not a rhetoric question).
> > 
> > Why not *what*?  Why not fork the file?  Why not apply this patch as-is?
> > Why not integrate this change in Automake?  A wee bit of redundancy can
> > sometimes help get over the noisy channel that is non-native language.
> 
> The patch itself looked okay; and since automake is not currently using
> flock, I guess I'm okay with forking for now.  Go ahead and apply it.

The patch doesn't constitute a fork from Automake.  It merely puts the
test in a file maintained by Autoconf, rather than where it would
rightly belong: in XFile.pm, maintained by Automake.

Pushed now; thanks.

Cheers,
Ralf