Re: [Wvware-devel] "wrapper" fixes for wvware-0.6.4

[Martin V. <mar...@hu...>]

On Sat, Mar 17, 2001 at 12:12:24PM -0700, Michal Jaegermann wrote:
> From: Michal Jaegermann <mi...@ha...>
> To: wvw...@li...
> X-Mailer: Mutt 0.95.5us
> Subject: [Wvware-devel] "wrapper" fixes for wvware-0.6.4
...
> Date: Sat, 17 Mar 2001 12:12:24 -0700
> 
> In my attempt to make vmware to do something more useful for me
                        ^^^^^^
That's another product ;-)
> I made various "peripheral" changes.  First I made sure that
> generated LaTeX files are using Type1 fonts as this makes quite
> a bit difference later on for Postscript and PDF.

Yes... for PDF-ing, the metafont fonts are a disaster.
Instead of times, you can also use ae, if you like to stick to Computer
Modern.
 
> I also rewrote various auxilliary scripts to make them safer
> (temporary files handling), more robust in a face of error conditions
> and not to leave junk around in failures, and to supply default
> output files if not specified explicitely.  Moreover a conversion
> to PDF works using 'pdflatex' instead of Adobe-only 'distill'.

Yes... I wonder why distill only? Best would be to support both. As
these are different pathways, perhaps the scripts should just have
different names.
 
> Modified scripts rely explicitely on "bash-ism".  Probably the biggest
> one is: 'name=${1%\.*}'.  If you prefer to keep these "bash-free",
> and some shells do not support such pattern substitution,
> then this construct can be replaced by something like that
> 
>     name=`echo $1 | sed 's/\.[^.][^.]*$//'`
> 
> and few instances of more readable '...=$(...)' can be made into backquotes.
> Probably the most troublesome is 'shopt -s nullglob' in make_epses.sh.
> A possible replacement is something along these lines:
> 
>    list=`echo ${1}*.wmf`
>    if [ "$list" != "${1}"'*.wmf' ] ; then
>       for i in $list ; do
>        ....
>       done
>    fi
> 
> but this may fail with some patterns (when there is a file $1'*.wmf', for
> example) although it should work well enough in practice.

This begs the 

QUESTION: Who amongst the users runs these scripts using a sh-type shell, but
has no possibility to use bash?

I know that AbiWord, an important "customer" of wv, runs on a heck of a
lot of platforms, but it calls wv as a library. So these scripts don't
matter there. I also know that wv is being run by some people on Windows
NT + CygWin. Does CygWin have a full fledged bash? What about BeOS, QNX?
Comments please?

> Originals were torn between using 'which' and 'type' to check for
> programs existence.  I changed them consistently to 'type'.  Actually
> 'type --path ...' would be even better but this is bash again.
> 
> A documentation was NOT modified to reflect that output file names
> are now allowed but not required.  This should be done if fixes, or
> something similar, would be accepted.

I can look at some of these, but Dom should look at the others. What I 
can judge looks good.
 
> Makefile for version 0.6.4 has some forgotten leftovers from previous
> versions.

Hmmm... datadir / libdir?

I see that the second part of the patch was already made in CVS.
i.e. placing libwv.a into $libdir. The CFLAGS part not. I believe you
are correct here, but I'll leave it up to Dom.
 
> I attach a tar.gz archive with all these patches.  It includes also
> an rpm spec file which is a modified spec from Red Hat Powertools
> distribution.  Use as you see fit.

Thanks! 
 
>   Regards,
>   Michal Jaegermann
>   mi...@ha...



-- 
Martin Vermeer  mar...@hu...
Helsinki University of Technology 
Department of Surveying
P.O. Box 1200, FIN-02015 HUT, Finland
:wq