From: Hubert Figuiere (hub@nyorp.abisource.com)
Date: Sun Apr 28 2002 - 05:43:20 EDT
----- Forwarded message from owner-abiword-dev@abisource.com -----
Sender: hp@icon.devel.redhat.com
To: "Tomas Frydrych" <tomas@frydrych.uklinux.net>
Cc: abiword-dev@abisource.com
Subject: Re: Pango portability (or rather the lack of it)
References: <3CC6DB11.5706.2C86B8@localhost> <3CC7D8CD.10199.5B7EC8@localhost>
From: Havoc Pennington <hp@redhat.com>
Date: 25 Apr 2002 22:52:08 -0400
In-Reply-To: <3CC7D8CD.10199.5B7EC8@localhost>
Message-ID: <y5wwuuv41ev.fsf@icon.devel.redhat.com>
Lines: 50
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.1
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
"Tomas Frydrych" <tomas@frydrych.uklinux.net> writes:
> Again, AW is _not_ a gtk-only application, so as long as gtk,
> win32, BeOS, QNX and Mac widgets do not behave the same way,
> we will end up with different behaviour in the editing window
> somewhere, whether we use Pango or not.
Maybe you need an abstraction layer with the ability to use Uniscribe?
;-)
> (1) I know of no such suitable renderer.
>
> (2) I have serious doubts, because of the GNOME-Pango
> alignment, that we can expect rapid development of Pango for any
> non-Unix platforms.
>
> (3) This is logically our best option, but our own development team
> is very small, and we have to consider this very carefully.
I understand. All I'm saying is, if you write from scratch you have to
do (3) anyway, so I don't really "get" why it's extra work to make an
existing thing portable vs. starting over entirely. GLib is certainly
designed to be portable and has been built on OSX, normally builds on
Windows, etc.
Paul suggested a mini-GLib, which may be a reasonable idea if
GLib is an issue.
Just to be sure it's clear, Pango has a GTK alignment, not a GNOME
alignment. GTK is committed to working on win32 as a design goal (and
would like to support Mac as well). This is different from
GNOME. Though the win32 port isn't great right now, it roughly works
and we are willing to take whatever patches are needed to make it
really good.
This is in fact a big reason I'd like AbiWord to use Pango, because
I'd like the win32 part to be worked on more intensively.
I do understand that it's always risky to use an external project, but
I also like to see "synergy" between open source projects.
> So, since you are in a better positon than any of us on this list,
> how much work would be required for us to get full Pango
> functionality with the FreeType2 backend?
Owen is much better qualified than I am to answer that. It seems to me
that the answer really requires understanding of both the Pango side
and the AbiWord side.
Havoc
----- End forwarded message -----
This archive was generated by hypermail 2.1.4 : Sun Apr 28 2002 - 05:43:21 EDT