Re: patch8.diff updated Was: Re: GCC's -fsplit-stack disturbing Mach's vm_allocate

[Samuel Thibault]

Svante Signell, le Wed 21 May 2014 09:47:08 +0200, a écrit :
> > > +# Special treatment of EWOULDBLOCK for GNU/Hurd
> > > +# /usr/include/bits/errno.h: #define EWOULDBLOCK EAGAIN
> > > +if egrep 'define EWOULDBLOCK EAGAIN' gen-sysinfo.go > /dev/null 2>&1; 
> > > then
> > > +  egrep '^const EWOULDBLOCK = Errno(_EWOULDBLOCK)' ${OUT} | \
> > > +    sed -i.bak -e 's/_EWOULDBLOCK/_EAGAIN/' ${OUT}
> > 
> > I don't understand why you both pass the output of egrep to sed, and you
> > give the -i option to sed. AIUI, the
> > egrep '^const EWOULDBLOCK = Errno(_EWOULDBLOCK)'
> > part is completely unused, so you can just drop it.
> 
> Well, the -i option is to get a backup copy for debugging purposes, can
> safely be removed.

Err, no, -i completely changes the behavior of sed, which then doesn't
read its stdin at all any more, it will modify ${OUT} in-place instead.
It happens that it is what you want here.  But then just drop the egrep
just before, it really is useless now. And drop the .bak suffix, I don't
think the gnugo maintainers will want to see a .bak file lying behind.

> Good luck making the patch better. It worked for me, but as I wrote
> before I'm no sed/grep expert.

I'm not the one to be convinced, gnugo maintainers are the one to be.
I'm just telling you in advance what *they* will tell you.

Samuel