Mostly pointless

The first is the thinking that people (in general) are going to write applications for GNOME specifically. They aren't. It's too niche to risk that, for both commercial and hobbyist developers. It restricts your potential userbase to a small niche (GNOME Linux user) of an already small niche (Linux desktop users). If you want to gain developers, you need to give them a reason to use your stuff, and until it's an acceptable cross-platform solution, it will be ignored. That's the basic reality. If I can write a GNOME application *and* have it work cross-platform, then you've given me a reason to consider using it. Until then, you've made a deliberate choice to limit who can use it, and by extension become a contributor to it.

On top of that, it's technically inferior to what else is readily available. So I have even less reason to consider using it. Even if you fixed the documentation issues, the bigger structural problems remain (and have done so since the start, in retrospect).

I'm lucky enough to work as a full-time free software developer (GPL and BSD libraries and programs for scientific and medical imaging); we're required to make our stuff run on all major operating systems. When time came to write a GUI frontend for one of my libraries, I went with Qt because it was realistically the only sensible option around for a cross-platform OpenGL image visualisation application. Its look and behaviour is completely native on all the major OSes and it works brilliantly. GTK+ doesn't offer that. If it did, it /might/ have been worthy of consideration. But it doesn't, and that means it's automatically excluded from many projects where it could potentially have been used. Until that is fixed, it will remain excluded in its niche. Your assertions about cross-platform applications are simply wrong, I'm afraid; Qt allows us to provide a first-class application on Linux and also on Windows/MacOS/BSDs. GTK+/GNOME would provide a mediocre application on Linux/BSD and a terrible application on Windows/MacOS, if it could be made to work at all. Your comment about "competitive advantage" is odd--both companies and individual developers would generally like to target as big an audience as possible; willingly excluding all but a tiny fraction of users makes no sense. In many situations, Qt allows provision of a Linux build where otherwise development would have been constrained to Windows/Mac; so in many ways is a positive enabler for getting Linux application support.

Regarding the community, yes it could be much better. Contrast it with others (Qt in particular), and there's a world of difference in the attidude and helpfulness of the participants. You know what the biggest difference is? Qt is actively used by many thousands of application developers, and the focus is on writing and delivering usable, functional applications. GTK+ and GNOME do not have that focus; they lost the people that mattered--the application developers--many years ago (I was one of them) and the level of discourse and current lack of direction reflect that.

Back when I was writing GTK+ applications commercially a decade ago, I submitted quite a few tested patches. Many sat unreviewed and unapplied in the GNOME bugzilla **for over a bloody decade**. That's the level of support a developer gets from the GNOME people: being completely ignored. While I was trying to write a free software editor application using the GNOME canvas, and providing patches to fix major problems with it, the GNOME community were busy creating >6 forks of it all of which were defective in different ways and all of which subsequently died. The result: today there is no canvas library (no, Cairo doesn't count since it's not a retained-mode system using the regular signal/event/object system), and GNOME kicked every developer who depended upon the old one. It was removed because it was "unmaintained", but if you don't have the resources to actually apply the patches to keep it going, that speaks volumes about the commitment there. They had the resources to create many incompatible forks, but not to apply a handful of patches?! And removing "old" but perfectly usable libraries just screws over those who are using them; way to kill backward compatibility and enrage all the users! GTK+ and GNOME have always treated third-party developers poorly, and the lack of good cross-platform support is a part of that, as is the lack of commitment to stable interfaces in the longer term. If they had, they would have retained some decent developers and it would still perhaps be a viable choice. As it is, they effectively alienated the very people who would have been their biggest supporters and contributors by marginalising and ignoring them. And the breakage with GTK+3 is just another set of problems on top of those which already existed long before then.

It's not too late for all these problems to be fixed, but it does require a drastic reorganisation of the project's priorities, and I'm not sure the rampant groupthink in the remaining developers will allow it.