State of GTK+ on Windows (Was: gtk 2 or 3)

[Neil Bird <neil fnxweb com>]

Around about 28/10/10 09:35, Tor Lillqvist typed ...
> GTK+ 2 is to some extent usable on Windows, sure. Unfortunately, for
> some aspects, earlier versions (up to 2.16 or so) are better than the
> later ones up to the current stable version (2.22).

  So what are the curent issues with the Win32 GTK+ port?


  We have a cross-platform C++ class framework that we've developed at 
work, but it still relies on Windows MFC for the main GUI aspects.  I've 
always anticipated that should we be able to move to a cross platform GUI as 
well as cross-platform core code (most of our systems are head-less Linux 
boxes), we'd want to plump for GTK+.

  While I am a fan of C++, I have issues with the Qt way of doing things: 
essentially, it's great if you want to start from scratch in C++, and do 
everything the Qt-way, with Qt classes from the get-go, but I had immense 
grief getting a simple proof-of-concept Qt GUI working [on Windows] on top 
of our classes, esp. with lockups due to threading issues, as I was using 
our threading and messaging classes (which we want to keep) and not the Qt ones.

  I had much better luck with binding to GTK+, which is better for us as 
it's cleaner to wrap your own C++ classes around basic C API (IMHO) than 
around other C++ classes;  it lets you map your ode to the API more easily.

  Ironically, I actually had no problems updating the GUI from the 'wrong' 
thread by using recursive mutexes for the GTK locks, and threading is the 
one thing GTK is not supposed to be great at under Windows;  Qt just locked 
up for no apparent reason, an *it* is supposed to be OK.

--
[neil fnx ~]# rm -f .signature
[neil fnx ~]# ls -l .signature
ls: .signature: No such file or directory
[neil fnx ~]# exit