This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
Re: gsltest problems
- To: wbrower at ll dot mit dot edu
- Subject: Re: gsltest problems
- From: Steve ROBBINS <stever at bic dot mni dot mcgill dot ca>
- Date: Thu, 20 Jul 2000 10:49:57 -0400 (EDT)
- cc: gsl-discuss at sources dot redhat dot com
On Thu, 20 Jul 2000, William Brower wrote:
> It is either a "sed" problem, or a regexp problem, or both.
> I can't help with either but can give you the evidence.
>
> The C program in the gsl.m4 contains symbols like $gsl_minor_version,
> etc. Those are replaced with the values obtained by running gsl-config
> and parsing the output with sed (around line 46 or so). I put in:
>
> echo $gsl_major_version
> echo $gsl_minor_version
> echo $gsl_micro_version
>
> to see what it thought those replacements should be.
> With GNU/Linux, I see
> 0
> 6
> 0
> as expected both by me and the C program.
>
> With that crummy SGI (not using GNU's sed), I see
> 0
> 0.6
> 0.6
[ ... ]
> I can send the SGI's sed man page if you think it would be
> helpful. Perhaps some regexp apologist can think of a more
> robust method?
I discovered the following through trial-and-error, so I'm an
experimentalist, not an apologist! ;-)
It turns out that SGI's 'sed' does not support the shorthand \? for
\{0,1\}. I changed the sed lines of gsl.m4 and got the right answer.
-S