Sure. The point was, if Sony waited until everything was perfect they would have fallen even further behind.

That's an interesting assumption. The switch from GTK+ to Qt isn't only a switch from a GUI toolkit to an expansive UI and application framework, though that is hugely beneficial, it's also intended to speed the development of both Maemo MeeGo itself and the applications that run on it.

Maemo hasn't just used GTK+; it has used GTK+ and Hildon. Up through Bora, the system used a forked, highly customized version of GTK+ 2.6. Going it alone on that got to be too taxing, and so with Chinook, Nokia broke backwards compatibility of Maemo and based Hildon 2.0 on GTK+ 2.10.

Again though, Nokia's finger friendly, touchscreen oriented changes weren't always accepted upstream, so they found themselves continuously having to adjust and reapply their patches to the rapidly evolving upstream GTK+ code. That work consumed resources that could otherwise have been used to move the OS forward.

Harmattan maintains much of the Gnome software stack, but switches to Qt as the primary toolkit to get away from the problem. Meanwhile, Nokia did contribute money to the Gnome Foundation to have Igalia work on integrating bits of Hildon into GTK+ upsteam for GTK+ 3. That should help with porting GTK+ 3 applications to MeeGo.

Is that one of the the reasons why Android was designed the way it was so that it would not have to worry about these dependency issues?

I don't follow Android closely, so I don't know how free it is of dependency issues. I do know Android made the news not long ago for not properly submitting the large number of Linux kernel customizations Google makes back upstream for inclusion in the mainline kernel.

Android's custom kernel code is publicly available, mind you. This isn't an issue of GPL violation. Like Nokia with GTK+ though, the further Google allows their fork of the Linux kernel to drift from the mainline, the harder it becomes for them to gain benefits from the improvements happening in the mainline.

The perils of open source software eh?

Forking isn't bad, but the ramifications shouldn't be taken lightly. Hence the focus in MeeGo of pushing all changes to the upstream projects and then benefiting when the improvements become available for all. Nokia, Intel, and the other companies involved have all burned themselves enough times in the past to know now that working upstream is the most sustainable method of participating in open source.

You can look at this with different points of view.

1_ What do we want as Nokia consumers, Maemo consumers, ignoring other platforms. What would make this "niche" more happy.
2_ What does Nokia has to do looking at all its consumers as a whole, and where they need to move inside the market to simply stay afloat, or to grow.
3_ What does Nokia has do to "tap the iPhone market", or how to release an "iPhone killer", etc.

These are all moving targets, with lots of uncertainties, but sometimes a few similarities... It is still very hard to do a "perfect decision" that would kill these three bunnies with a single whack.

Apple has a huge cult following, and were coming from the big successes with the iMac then the first iPod. They kind of intruded into the mobile market bringing along its cult, and ripping off lots of ideas that they weren't really the first to have. I am not saying this is "unfair" or anything, just that it's not as simple as having the cool and innovative idea. There is a lot between that and effectively making the product a success. Even harder to have a success of sales _and_ of media.

You talked a lot about iphone being "sexy", while Nokia's would probably be "unsexy", but what exactly does that mean?... And how do you build _that_??

Tomi Ahonen wrote a huge (as always) blog post the other day, saying Elop is delusional, and he mentions a fun fact. In Japan it seems very few people have an iphone as their main phone. They have other more advanced ones, fit to their Japanese needs, but buy iphones just as a cool piece of fashion or whatever. Just because it's a "sexy" trinket. It's hard for Nokia to worry not just about how to make a great phone, but also how to make it become "sexy" like this. Maybe an impossible task.

What are Nokia's assets?... Nokia has, first and foremost, a great design team, and great engineers to build the hardware and antennas, etc. They are also very "globalized", caring a lot for having multiple products to fit multiple, specific needs...

Myself, I find the top Nokia phones more "sexy" than any other brand at fist sight. I really cannot think of having any other. The iphone looks to me just like a broken rear mirror from a bike, or something. Jessie's Girl, she is the sexy one to me.

Now, what are we talking about again?...

OK, Qt... Qt seems to me to be a great strategy for both point-of-views 1 and 2. I don't know about 3, but this is probably unrelated and unsolvable.

Nokia can't make anything to "become" Apple. Their consumers and the media _want_ Steve Jobs, specifically. They want the little Apple. You really can't do anything. What they can do is, in more or less Elop's words, ride the next "wave of disruption". Having good "cloud" services is probably one great strategy. Nokia did try to create lots of services, but it seems it didn go too well. Some will probably say it was "too early" some day. Unless they do it right, and right now. Qt or no Qt.

Or, I don't know, they could spend some extra time to make their patches acceptable and stop having to maintain them externally.

It was just a lucky break that there was an alternative upstream toolkit available to buy and switch to. But it did come at a high cost - Fremantle was half-finished because they didn't want to spend too much effort developing a dead-end platform, and Harmattan still hasn't seen the light of day even in SDK form 2 years after the switch announcement and counting.

Since there is a lot of discussion about Qt, I think this might be a good thread to ask about this:
-Nokia and Microsoft have announced that Qt will not be available for WP7. But why? Couldn´t Nokia push it there by themselves? Part of the MS/Nokia deal was that Nokia can customize WP7 (something that other WP7 OEM's are not allowed to do). For example, I would assume that Nokia probably wants to use some other processors than just Qualcomm ones (which are the only ones supported by WP7 at the moment) in the future to push the hardware costs down and most probably Nokia has to do the hardware adaptation themselves.

So is the situation really that Nokia can't make Qt work on WP7 or is it that they for some reason don't want to do it?

(just thinking out loud, hope this makes any sense)

You've already answered. Lord Vader would not appreciate it.