On Mon, Nov 20, 2000 at 11:18:09AM +0100, Joaquin Cuenca Abela wrote: > Sam TH wrote: > > > On Sat, Nov 18, 2000 at 01:19:02PM -0500, Miguel de Icaza wrote: > > > > > > Havoc has presented a lucid explanation of the issues on XP toolkits. > > > Any chance we can get Abi to use Gtk+ as the main toolkit, and > > > encourage ports by having the underlying Gtk+ ported over your > > > favorite platform? > > > > A while back, Abi was ported to GTK+ on Windows. I don't know if that > > still works, as no one uses it currently. But it is certainly > > feasible. We would welcome people providing the code to allow GTK+ > > builds on more platforms than we have currently. > > no, if I remember correctly Abi was ported to GTK+ on Windows *trough* a > windows X server (ie, the port didn't use the gtk+ port to windows, but the > cygwin stuff + a X server). Yes, your memory is correct. See http://www.abisource.com/mailinglists/abiword-dev/00/March/0267.html for more details. However, a port to GTK/win32 would be welcomed. It would probably require some weird changes, but wouldn't be that much code. > > > However, it seems like you are asking us to quite providing native > > front ends, and only use GTK+. Of that, there is no chance. > > Just wondering, Sam, do you have any technical reason? > > Havoc has point out a problem with the current design. We're getting more > "featured" in the gnome-frontend that in the {Windows,BeOS,QNX}-frontend. > For instance, now AbiWord can export to pdf only in the gnome-frontend, it > has nice icons in the menu items only in the gnome-frontend, the toolbar can > be placed vertically only in the gnome-frontend, it has a "print preview" > only in the gnome-frontend, etc. > It should be noted that the GTK front-end doesn't have these features either. However, we have these features because we got them basically for free. GNOME-print, which provides most of the user visible current gnome front end benifits, wasn't written for AbiWord. Dom, with help from Chema and Lauris, just allowed us to *use* it. Now, lots of other platforms could have neat features like this. I'm sure Windows has something. But they aren't free software, and they aren't as easily available. However, if someone were to do a KDE port, then AbiWord could take advantage of all these features in KDE. The problem isn't with Abi's design, the "problem" is that GNOME has produced so much cool free stuff. > If we want to have all these features in the others frontends, we must to > code it one time per toolkit. I think that it will be smarter to just code > to a multiplatform toolkit, and achieve the platform specific look with a > theme. Swing shows beautifully what can be achieved (just multiply by 10 > his speed, and improve a bit his "windows" theme, and you get a killer > multiplatform toolkit). > > Of course, I'm speaking of a post 1.0 change. And before all that becomes > possible, we need: > > 1) "acceptable" ports of gtk+ to windows, BeOS and QNX (the Abi supported > platforms) And Mac OS X (for which a port is currently in progress). > 2) "acceptable" themes for each platform Remember what people said (and still say) about Mozilla? That it looks horrible, since it doesn't use native widgets. People said that less on Linux, since everything does really look different. But Windows people, and especially Mac people, hated it, and many probably still do. So GTK will have to look *really* good, to a degree that would surprise me, to look as good as, say, Windows Me native widget, and Mac OS X native widgets, and QNX Photon native widgets. > 3) idem for gnome (if we want to use the gnome platform) I've already addressed why it's important to maintain GTK seperate from GNOME in another email. > 4) the change has to be accepted by the current Abiword developers > (specially interesting will be the opinion of Thomas --which ported AbiWord > to QNX and BeOS--) This is a big problem. Our windows developers are just that, windows developers. Most likely, they've never used GTK. Why should we make the barrier to entry even higher. We have basically one Be and one QNX developer each, and just a handful of Windows developers. We need to make this harder for them like we need a hole in the head. > 5) work, work, work > > So, what do you think? what are you're reasons again this change? The single most important reason that AbiWord will be successful is that it is XP. Even more important than that it is Free. And to as XP as we are (which is more than just about any other GUI app) we need to be easy to install and easy to develop for on lots of platforms. That means using the toolkits developers are used to. That means not having to ship all of GTK and GNOME with our 3 MB binary. (This will, of course, be different, when Windows 2005 includes GTK. Ha.) There are lots of things about AbiWord that should be changed (like all the bugs). Our cross-platform nature isn't one of them. sam th sam uchicago edu http://www.abisource.com/~sam/ GnuPG Key: http://www.abisource.com/~sam/key
Attachment:
pgpaY0iS6Nfky.pgp
Description: PGP signature