Re: [bug-gv] Working around gs issues by changing to a safe working dir

[Markus Steinborn]

Hi Bernhard,

Bernhard R. Link schrieb:
> Attached patch is not yet in upstreamable quality, but I thought I
> better already post it here in case someone has comments or want
> to improve it before I have the time.
>    
Thanks.
> The idea is to avoid any working directory problems by running gs
> only in a safe working directory.
>    
Sounds reasonable, as we have to deal with broken upstream versions. In 
between I managed to find the broken upstream commit: 8954. That means 
that all upstream releases from 8.64 up to now are broken, i. e. "-P-" 
does not work as expected.

I'm just wondering why debian lenny also has this bug... I am running 
GPL gs 8.62 without having "-P-" broken.
> The patch is quite minimal and targeted at the Debian packages,
> so still todo:
>
> - modify Makefile.am instead of .in
>    
done and ported to git master (untested und not committed). Before doing 
the other parts, I#d like to do some tests, you see: Latest RHEL is 
shipped with Ghostscript 8.14, so we need to test really old ghostscript 
versions.
> - create the directory on install time
>    
Well, personally I like the option being able to test GNU gv without 
having it installed, just "configure && make".
> - port to non-glibc (currently uses canonicalize_file_name)
>    
Surely an important point.
> - documentation updates
>    
I agree.
> - perhaps a command line argument.
>    
I'll do a command line argument (and the counterpart --no...) after 
ghostscript compatibility tests. They are more urgent.