John Spray wrote:
On Wed, 2006-06-28 at 10:31 +0200, Jean-Marc Lasgouttes wrote:
The problem as I see it is that people making changes to the core don't want to have to make corresponding changes to three frontends -- it ties their hands. They're right. I think it's a real drag. It's prevented me from fixing bugs in the controller classes when I found them, because I would have to change all the other frontends at the same time. So, here is what I suggest: if core developers don't want to make the occasional few-line change to keep a frontend working, fine. However, any time a change is made in the qt frontend to accomodate a backend change, this should be dutifully noted somewhere (on the wiki?) so that at some point when someone (probably me) wants to bring the gtk frontend back up to speed, he can. Also, at such times, it should be asked "Why are we having to change a frontend? Why is this not abstracted into a controller class?"
Agreed.
If gtk and xforms are eliminated, then only qt remains (I don't care which version, neither do users). At this point, the abstraction of frontends will inevitably decay. There is merit in the argument that having multiple frontends, even if not used, "keeps us honest". Perhaps that could be the right decision, to leave the multi-frontend experiment behind. But if that is the case, it's a decision to be made on its own, it should not happen as a side-effect of trying to reduce maintenance overhead.
I disagree with you here. You don't need to have multiple frontends to have a clean kernel/gui separation. Now that we know how to do it, it is really easy to draw a line between the two worlds. This line is still dashed right now and more work is needed to provide this clean separation.
Executive summary: - If you want to just have one frontend, say so -- don't just imply it by getting rid of the allegedly troublesome others.
I am just advocating that only one frontend should be required to compile and work up until this separation is done properly. Then it will be even easier to make a new frontend. I never said and will never say that the other frontends have to be erased from svn. IMHO, even xforms should stay for a while. If someone is willing to do the cleanup work for xforms he/she should feel free to do so.
That said, I personally don't think it is useful to have multiple frontends for a given platform but there might be some other advantages for sure. The gtk frontend could well be useful for an embedded platform that supports only GTK for example.
- If multiple frontends are to be maintained, then even a stalled lyx-gtk is useful for attracting gtk developers,
Agreed.
and for maintaining the frontend-independence code.
Disagree ;-) Abdel.
