Re: [gtkmm] porting projects from gtkmm-1.2 to gtkmm2

[Leandro Fanzone <leandro hasar com>]

It's not an exaggeration, my application's basement was practically a
widget that dissapeared completely, and replaced by a totally different
and very complex one. I said that my case perhaps it's not the common
case. I just wanted to tell my case. 
At the time the application was finished and running, there was no hint
that CList would die any time soon (Karl's times). I was forced to port
it when the problems with anti aliased fonts arrived. We even evaluated
if it was worth to port it to gtkmm 2.0 or to QT, facing a new rewrite.
The strong argument was the chance of API breakage of this magnitude
could happen again. It was me who decided to stay with gtkmm, in spite
of the hard work. I am talking about a 15,000 lines program, that took
a month and a half to port it and get it working again.
In any case, I have a simple sed script to change the obvious things
(includes, connects, etc.), if it can be of any use. 

Leandro.

Murray Cumming wrote:
> On Mon, 2003-01-13 at 18:09, Leandro Fanzone wrote:
> > My own experience is that I almost had to rewrite everything from 
> > scratch. My application was heavy reliant on Gtk::CList which doesn't 
> > exist anymore, for example. To mimic all the things I had done in CList 
> > with TreeView was a headache after another. Also all the fonts 
> > management changed radically. I began to port module by module, until I 
> > realized that I should better begin from scratch, reusing some of my old 
> > code here and there, but thinking on the new platform from ground zero.
> > Well, that's my case, I think in other cases it could be far more 
> > easier, naturally.
> >   
>
> I think that's an exaggeration. I have ported a large app (Glom), and
> all of the gtkmm examples. It was very straightforward. Code that uses
> CList or CTree needs to be almost completely rewritten. But you should
> have known for a long time that those widgets would die.