[Mono-list] Linux.NET

[Paolo Molaro]

On 04/13/02 Dan Lewis wrote:
> Is there a case for ditching glib? We use barely a quarter of it (see attached
> file for a simple function breakdown). The data structures and string routines
> are probably the most useful - Paolo's hashtable patch shows that they can be
> lifted out safely. Assertions and reporting could similarly be lifted.
> 
> GLib unicode support seems a little awkward for Mono's needs -- we could really
> do with our own manipulation functions (I'm just finishing off some C# encoding
> implementations at the moment :) And memory management eventually needs to be
> closer to the garbage collector.
> 
> That leaves a few system functions -- most notably gmodule.

GLib has been a tough dependency to have for systems different from
Linux, simply because it was in development and had build issues.
I hope that now that GLib 2.0 is out, there will be fewer issues,
especially if people working on the 'weird' platforms submit patches to
the GLib maintainers for build problems or other portability issues.

I didn't copy the hash code lightly: it was the only way to get
functionality GLib doesn't provide while keeping some API compatibility
and I would consider a mistake getting rid of the GLib dependency.
I may also try to get a patch in GLib that allows us to get rid of
the copy.
After all, you're going to have quite a bit more problems porting mono
than porting GLib and porting GLib makes more people happier.
If there are issues porting GLib, you'd have the same issues porting
our replacement for the GLib functionality, you'd just duplicate code
and efforts.
If the GLib mainatiners won't accept patches in a timely manner, that
could be a different problem, but I don't think that is the issue.
Serge: what problems did you have? How can we help you?

lupus

-- 
-----------------------------------------------------------------
lupus@debian.org                                     debian/rules
lupus@ximian.com                             Monkeys do it better