Otte: staring into the abyss [LWN.net]

Otte: staring into the abyss

Posted Jul 27, 2012 17:56 UTC (Fri) by raven667 (subscriber, #5198) [Link]

Right, that's its own abyss right there.

The development efficiency gains, which are likely entirely illusionary, don't offset the trashing of an entire software ecosystem. There are plenty of examples close at hand.

Otte: staring into the abyss

Posted Jul 27, 2012 18:15 UTC (Fri) by boudewijn (subscriber, #14185) [Link]

Rewriting stuff from scratch is seldom the answer indeed... Which makes it worse if it looks like the current platform these great apps are written on depend on the work of one, increasingly frustrated person. I know that, for instance, the gimp people are working on GTK3 support, but they also have to contend with really bad bugs and lack of support for Windows and OSX. I don't like that situation one bit. Porting gimp to Qt was a demonstrated possibility way back in the closing decade of the last century, I doubt the feasibility of that at the moment.

Otte: staring into the abyss

Posted Jul 28, 2012 22:30 UTC (Sat) by pboddie (subscriber, #50784) [Link]

I know that, for instance, the gimp people are working on GTK3 support, but they also have to contend with really bad bugs and lack of support for Windows and OSX.

I imagine that things have moved backwards in the last few years, then. For a long time people complained about Windows and OS X support, and for a short while it looked like all the pieces were in place, but now we might be back where we were before: so much for that "ISV friendly" weak-copyleft-tending-towards-permissive licensing that meant that anyone doing the portability grunt-work was likely to keep it to themselves rather than share and collaborate; there's a danger that Qt could go the same way if various companies write off their investments.

It's all very sad and a reflection of the last five years or so in the wider free desktop scene.

Otte: staring into the abyss

Posted Jul 29, 2012 12:50 UTC (Sun) by Frej (subscriber, #4165) [Link]

GTK3 just got in homebrew that's what actually matters on OSX? :)

And GTK3 works better with the quartz backend than gtk2 ever did, i think the loadable backends helps quite a bit here.

Otte: staring into the abyss

Posted Jul 28, 2012 4:28 UTC (Sat) by daniel (subscriber, #3181) [Link]

"Rewriting everything in order to just accomplish what it already does doesn't seem like a good path forward."

Who said anything about rewriting everything? For one thing, the entire back end should stay unchanged, and if it can't because it lived its entire life so far joined at the hip to its GUI then that needed fixing anyway. Fix any places like that as the first step of the rewrite while still staying with GTK. That is normally a fairly easy iterative process.

Then... enjoy the ease and rapid development of QT. It's just really easy and pleasant, speaking from experience. And keep the GTK interface around as a compilation option so long as it is not too much of a burden.

In any case, the basic structure of the interface will not change because QT follows essentially the same model (register a widget; register layouts in the widget; register subwidgets in the layouts; connect UI events to program actions). Except easier to write and easier to debug, easier to get the defaults right, and easier to add new functionality in a generic way.

Otte: staring into the abyss

Posted Jul 28, 2012 7:55 UTC (Sat) by boudewijn (subscriber, #14185) [Link]

"Who said anything about rewriting everything?"

Drag did.

In any case, while there might be some way to do a semi-automated port (KDAB has tools for that, they used to port lots of code from Motif to Qt), I don't consider such a scenario as realistic. You can quibble about the difference between "rewriting everything" or "just adding another front-end", but the fact remains: it's a nearly impossible amount of work.

I'm not sure Raven667 got the point, though -- he talks about "the development efficiency gains, which are likely entirely illusionary". The issue isn't that these huge and important projects would enjoy the benefits of working with Qt, it's that they don't die because their platform is starving to death.