Re: pangox.aliases on various X platforms...

[Havoc Pennington <hp redhat com>]

Arnaud Charlet <charlet ACT-Europe FR> writes: 
> But this is definitely an issue when doing binary packages and when
> distributing these binaries, since people may or may not have the exact same
> configuration. How is this supposed to work ? What is the exact use of
> pangox.aliases and what are the consequences of an innacurate file ?
> 
> What would be a way to generate a proper pangox.aliases ?

It can't be autogenerated; otherwise it wouldn't be necessary.

The point of this file is to let people specify fonts that look good
and are available. If you want to create a generic pangox.aliases that
always works, you can just put in "fixed" for all the fonts, or "*",
or equivalent, right. But the results will be plenty unattractive.

> I hope you realize such dependencies makes it even more difficult to create
> self contained binaries for platforms that do not have Gtk+ pre installed
> and pre configured (the usual case on most unix and windows platforms).
> 
> Currently with Gtk+ 2.0, you need to distribute about 3 or 4 files, and set
> about the same number of environment variables to distribute something based
> on Gtk+ 2.0, compared to Gtk 1.2 where one could distribute a single
> executable.

I think we do realize that, but it's something we'll need patches or
concrete suggestions for. There are 400-something open bugs and Owen
is badly backlogged.

Havoc