On Tue, 2002-02-26 at 16:20, sam th wrote: > On Tue, 2002-02-26 at 14:12, Leonard Rosenthol wrote: > > At 6:58 PM +0000 2/26/02, Andrew Dunbar wrote: > > > >Also I'm wondering if this decision should be taken > > >seperately from deciding whether we should use Pango > > >at some point. > > > > Pango, as cool as it is, requires a LOT of Glib/GNOME to > > function and that's problematic on our full platform list. > > > > Actually, Pango just requires GLib, which should work on all of our > supported platforms. It definitiely works on Win32, *nix, and is being > ported to BeOS. (www.gtk.org/beos) And, I remember Havoc Pennington (a > GTK+ maintainer) saying that he would like it to be as portable as > AbiWord. >
Pango uses Glib2, which has other dependencies and idiosyncracies. wv and NOT Abi uses glib1.2 Glib2.0 uses: * threads (win32, posix, solaris, NSPR) including mutexes and conditionals * the gobject system * libiconv * a simple xml parser (i believe) And a bunch of other stuff that we either already provide or have workarounds for. Plus it has lots of baggage that we don't need (string tokenizer class, hashtables, ...) or don't want. Ideally GLib2.0 could be ported to QNX, BeOS, MacOS, ... with a minimum of work, but it'll probably be non-trivial for at least one of those platforms. Add that to the fact that I'm not convinced that Pango is the best solution for us post 1.0, and you get my resounding "no" opinion on Glib2.0 stuff, at least for now. Freetype is very simple to use, highly portable, and solves a very centralized and small (yet very important) problem of ours. Using Pango will undoubtedly require a MAJOR overhaul of BOTH our and Pango's internals and is something that hasn't been duly researched into for our purposes. IMO, using the word "Pango" on this list at this point is just another buzzword. Dom
signature.asc
Description: This is a digitally signed message part