Re: [Quilt-dev] Re: config-cc patch

[John Vandenberg]

On 8/16/05, Dean Roehrich <address@hidden> wrote:
> On Mon, 2005-08-15 at 16:20, John Vandenberg wrote:
> > On 8/16/05, Dean Roehrich <address@hidden> wrote:
> > > On Fri, 2005-08-12 at 21:52, John Vandenberg wrote:
> > >
> > > >
> > > > In order for quilt to work on Solaris, the patch sun-backup-files.diff
> > > > is also needed.
> > >
> > > Sorry, I missed something here.  What changes allowed quilt to work on 
> > > Solaris?
> >
> > Hi Dean,
> >
> > I have a patch series that supports Windows, SuSE 7.2, Solaris, and
> > recently also OS X.
> >
> > http://zeroj.hda0.net/quilt-patches/
> 
> Thanks, I'll go through them.
>
> > > I have a small collection of my own changes that I've been meaning to 
> > > pass through
> > > the mailing list soon.  I have to do things like change the chmod in the 
> > > Makefile.in
> > > to not use the --reference arg; change patchfns.in to not use -n on the 
> > > tail command
> > > int top_patch(), and there's a change in a find command later.  I have a 
> > > change
> > > in pop.in to use a little perl inline script in place of the tac command. 
> > >  I think I
> > > have a few other things in there, too.
> >
> > I haven't noticed the tail -n problem yet, and there are probably
> > others we have each missed.
> 
> Are you using /usr/xpg4/bin/tail then?
> The /usr/bin/tail on Solaris 10 doesn't have -n.  This -n happens in the 
> top_patch()
> function.
> 
> Hmm, I wonder what other goodies I'll find in /usr/xpg4?

Without being aware of it, I was using the tail from GNU coreutils. 
xpg4 looks like a haven for all the good stuff; I'll have to add that
to my path.  Thanks.

quilt also depends on "cp -l " from fileutils; any chance Solaris has
a suitable binary hidden away somewhere?

I have added a TODO item to add a check for a working tail -n, and
maybe even write a test case.  All the tests pass, so if anything is
broken, its because it lacks test coverage.

I wonder what other small glitches are hidden away.  I dont think
diffstat is covered in the tests; any other tests that need to be
written?
 
> > Which getopt did you use?  I wrote a perl version of getopt because
> > most of the platforms that we develop for don't have a suitable
> > version.
> 
> I ported the one from util-linux-2.12q :)  Brute force...icky.

haha.  I ported the very same beast to Windows using MinGW, so I feel
your pain.  tac and column also reside in there, so it would be ideal
to have that package on all platforms.  I decided to write a perl
version because I couldnt see util-linux adding support for Windows,
OS X and Solaris in a hurry.  As it turns out, util-linux may be going
in that direction with the latest auto-confisication.

http://kerneltrap.org/mailarchive/1/message/101522/thread

> I like a lot of the things I see in your config-misc.diff.  The getopt.in is 
> worth
> arguing over, but the autoconf-related @CMD@ pieces shouldn't be a problem.  
> I think
> your tac(1) substitute shouldn't need the map function, though.

thanks.  I've updated the patch.
--
John