Re: Patches for OS/2 (2)

[Andreas Buening]

Eli Zaretskii wrote:
> 
> On Wed, 13 Nov 2002, Andreas Buening wrote:

[snip]

> > > DOS and OS/2 maintain a notion of "current drive". The
> > > effect that you achieve through UNIXROOT=f: is the same
> > > as the user would obtain through the command "cd f:".
> > > The handling of pathnames without drive letters is
> > > already handled inside EMX. I see no reason to duplicate
> > > it in the software above it.
> >
> > No. This would imply that the software only works on the drive
> > it is installed.
> 
> How about using one of the following solutions (or both of them)?
> 
>  1) Make libintl look for its files in some directory relative to the
>     place where the executable program is installed.  For example,
>     if foo.exe is in f:/usr/local/bin, make it so libintl looks for its
>     files in ../share/ relative to where foo.exe lives.  This could be
>     implemented by making foo.exe push into environment some variable
>     that libintl would look at, and making f:/usr/local/share the value
>     of that variable. (foo.exe knows where it lives by looking at its
>     argv[0].)

This is much more complicated because
a) If libintl has to look for an environment variable, why not UNIXROOT?
b) The source code of every executable that uses libintl would have to
   be modified.
c) I'm not sure about this but is it guaranteed that argv[0] contains
   a real path and not just "foo"?


>  2) Is there some environment variable used by EMX to point to the root
>     of its installation?  If so, you could use that place as the
>     replacement for /usr/local, and configure libintl look in its `share'
>     subdirectory.  In other words, assuming that, say, $EMX_PREFIX points
>     to the root of the EMX installation, make libintl look for its files
>     in $EMX_PREFIX/share/.  That is, configure libintl with
>     --prefix='$EMX_PREFIX'.

Not, there is no such variable. And there is another problem. If the
user compiles a package with --prefix=x:/this/is/my/prefix then it
must be guaranteed that this is not overridden. The user must have
the last word.


Bye,
Andreas