| |||||||||||||
Dricot: A freasy future for GNOMEDricot: A freasy future for GNOMEPosted Aug 11, 2012 13:01 UTC (Sat) by rleigh (subscriber, #14622)In reply to: Dricot: A freasy future for GNOME by jospoortvliet Parent article: Dricot: A freasy future for GNOME "Pick the right tool for the job, I'd say." The thing is, GNOME2 was the right tool, and GNOME3 is not. GNOME2 has a massive userbase--it was the default desktop on Linux for a decade after all--which has been discarded on a whim. The vast majority of GNOME2 users are on desktops and laptops. Why drop them? There's precious little evidence that of the tiny tablet market, any of them would want to use GNOME, so why chase after that minute niche rather than catering to your entire, massive, desktop userbase. There's a good reason why there's a great deal of displeasure over GNOME3, and it's entirely rational. Hundreds of thousands, if not millions, of people use GNOME2 every day at home and work. And they've been told categorically that they don't matter, and won't be catered for. "I think it'll do well at Point-Of-Sales and Kiosk systems" GNOME2 was already adequate for such systems, if not overkill. Back in the 2004-5 timeframe, my full time job was working on prototyping a touch-screen Point-Of-Sale system using GTK+, on GNOME. These are highly specialist custom applications. We used custom touch-friendly entry widgets. This logic was in the application, not the desktop environment. Such applications typically run fullscreen without actually using a desktop at all--they are generally locked down custom appliances, not general purpose. If the existing userbase is being discarded in favour of this... it's a completely wasted effort. And it's also a tiny niche market--certainly nothing to abandon the existing userbase over. The above PoS effort was ultimately abandoned. The GTK+ and GNOME libraries and bindings were not of good enough quality to make it a realistic proposition. And this is still the case today. A GNOME "OS" or "SDK" is completely unrealistic until there's some evidence that stable, usable, not horrendously buggy library APIs can be maintained by the GNOME developers. That's not happened in the last decade, so it's not a realistic expectation for that to improve any time soon. Merely wishing it does not make it so--and given the poor maintenance of the existing codebase, it would be foolish to commit it using it.
(Log in to post comments)
Dricot: A freasy future for GNOME Posted Aug 11, 2012 17:28 UTC (Sat) by hummassa (subscriber, #307) [Link]
> GNOME2 has a massive userbase--it was the default desktop on Linux for a decade after all
That made me laugh. Hard.
AOL Posted Aug 12, 2012 17:29 UTC (Sun) by Wol (guest, #4433) [Link]
Likewise the comment about KDE not being used ...
Okay, I think SuSE is about the only distro that defaults to KDE, but Kubuntu certainly seems to be popular.
But I'll add to that, my *limited* experience is that nobody uses Gnome!
I've now installed xfce and lxde on my system, but mostly because I've got fed up with dealing with the fallout of "emerge -u kde" in a kde console window. (The update takes out konsole, which takes out emerge, which leaves me with a mess of an upgrade to sort out!)
I've always been a kde guy, even when nepomuk was enabled by default and had a habit of killing systems... (as somebody pointed out elsewhere, even if it was infuriating, the kde devs did say "we're going to fix it" rather than what I see reported of Gnome guys saying "live with it").
Cheers,
AOL Posted Aug 12, 2012 20:05 UTC (Sun) by nybble41 (subscriber, #55106) [Link]
> I've now installed xfce and lxde on my system, but mostly because I've got fed up with dealing with the fallout of "emerge -u kde" in a kde console window.
screen emerge -u kde
Dricot: A freasy future for GNOME Posted Sep 6, 2012 9:08 UTC (Thu) by jospoortvliet (subscriber, #33164) [Link]
Ok, talking about SDK and software development - GTK is lightyears behind everything else. Qt is currently clearly the industry-standard for stuff like POS, info screens, custom touch devices and all that, with or without KDE stuff on top. So GTK was the WRONG tool for THAT job, by a huge margin I'd say.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 12:29 UTC (Thu) by rleigh (subscriber, #14622) [Link]
The decision was not clear-cut back in 2004, and even now, I'm sure that the GTK+ developers would contest this. You also need to consider that back then, people were adapting GTK+ for use on embedded devices, including touch sensitive and more specialised device-specific input. Even now, these things are still possible.
The problem with all UI libraries is that they tend to become their own enclosed universe, which you have to buy into completely. GTK+ suffers from this with GObject and the whole consequences of that. Qt suffers from its complete aversion to the STL--I might want to use it without all its non-standard types. And others like FLTK have a backward and non-extensible widget model by putting all the configurable properties in the base class. I've yet to find one which is just usable without all the extra junk; the GTK+ widget packing model is quite elegant, so it's a shame about the rest.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 12:46 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]
Technically, Qt seems to have a number of advantages. Undoubtedly it has a lot more full time developers working on it. However from a governance stand point, GTK is better. It is managed by a non-profit and doesn't require anyone to sign agreements before accepting contributions. Also Qt seems to be under turmoil now after Nokia abandoned Linux development (which to be fair, also has affected GTK) and it is not clear how much of a good free software citizen Digia is going to be. If Qt would be moved to management under a non-profit foundation, that would be a great outcome.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 13:34 UTC (Thu) by jospoortvliet (subscriber, #33164) [Link]
Yes, the future is uncertain, that much is clear. Then again, with still orders of magnitude more hackers on Qt (even if Digia would walk away, Qt would still have 10 times as many full-time hackers than GTK) and with many large companies like IBM, BMW and most of the movie industry all depending on Qt - I doubt it'll go the way of the dodo.
The governance thing I agree with, I would prefer a foundation too. Then again, a company can probably put in more engineers than a foundation could - I doubt even a hugely successful Qt foundation would manage to employ 200 developers. Realistic would be more like 20 and that'd be a huge decrease from the current situation.
And again, these numbers (both users and contributors) are so far ahead of any other FOSS toolkit I think we can safely say there is one de-facto standard toolkit on Linux and it's very Cute ehrm Qt ;-)
Dricot: A freasy future for GNOME Posted Sep 6, 2012 13:46 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]
I am not sure if Digia walks away from Qt, there will be a lot of full time developers left. Being managed by a non-profit doesnt preclude commercial participation in any way. It merely provides a level playing field and can very well *increase* commercial participation. Ex: Eclipse.
The claim of one defacto toolkit is bogus and you are very much overselling it considering the presence of desktop environments like GNOME and Xfce not to mention a fairly large number of third party and ISV applications. Ex: VMWare, Adobe etc.
Toolkits in Linux is not even limited to merely Gtk and Qt either. Firefox and Libreoffice for instance have their own toolkits but imitate GTK look and feel. There are a odd few apps using WxWidgets, Fltk and so on. This comes at a very significant cost. Note when Libreoffice finally got font anti-aliasing.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 15:36 UTC (Thu) by jospoortvliet (subscriber, #33164) [Link]
I know that a non-profit doesn't preclude commercial participation - I'd even argue the opposite. I just would not expect a non-profit for Qt to be able to get enough funds together (or enough help from commercial parties) to hire 200+ people. I heard rumors about a Qt foundation but the # of developers in that would've been at least 10 times less than the ~250 now.
Anyway, those talks are suspended until it's more clear what Digia is going to do - for now, at least they talk the right talk. Let's see if the walk it, too.
About the defacto toolkit - the fact that there are plenty legacy apps still using GTK doesn't mean it is anything but a legacy toolkit. It doesn't have the industry cloud Qt has - heck, where do you think the money to pay 250 full time developers comes from? Much of those companies are of course not visible to the consumer - the devices are often not identifiable as running Linux, let alone Qt (In Vehicle Entertainment!) - or you never see them (movie industry, medical industry etc). But they bring in enough funds to make quite a number of consulting companies quite profitable - Digia is merely one of them and wasn't even the biggest until they made the deal with Nokia and now bought Qt.
So, in the visible, fanboy-dominated FOSS world, GTK is still a major player. Just like KDE and GNOME matter for the FOSS world. But outside of that its not a major toolkit in any way, just like nobody knows KDE (or GNOME). With Samsung and Intel behind it I'd even bet EFL has more cloud (developers aware of it; developers able to use it; etc) than GTK these days and if it doesn't it soon will.
And even IN the FOSS world you see companies like Canonical moving to Qt development for their new projects and long-time GTK-only consulting companies doing more and more with Qt. And it's not a bad thing, it's good - Qt is a good toolkit, great to work with, has a good reputation in the non-FOSS world too, lots of skilled developers - and it's fully free and open. What more do you want than popular, successful, high quality, well documented, LGPL and with open governance? Oh, and a deal with the KDE e.V. to protect the long-term interests. All good, in my book...
It is good to see the FOSS world rally behind something like we're (almost) all behind Linux as a kernel, glibc, GCC Apache etc etc. Not that there are no alternatives - some alternatives trying to keep the big projects honest is good. Makes it easier to get rid of the big projects if they screw up, gives new ideas a chance, etcetera. But having a clear, prominent winner also creates clarity for newcomers and third party developers. So I think it's good...
Dricot: A freasy future for GNOME Posted Sep 6, 2012 16:32 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]
"I just would not expect a non-profit for Qt to be able to get enough funds together"
Maybe but that is irrelevant. The function of the non-profit isn't to raise funds primarily. Linux kernel doesn't depend on funds from Linux Foundation nor does Eclipse depend on Eclipse foundation raising funds directly. Why should Qt be any different?
"the fact that there are plenty legacy apps still using GTK doesn't mean it is anything but a legacy toolkit"
This is a good example of a circular argument. Plenty of companies outside of the FOSS world do *new* development in GTK as well. Heck, I just talked to one writing a accounting app in GTK. So I consider your argument very weak and bogus.
The reason Qt got funding is because of the dual licensing model. I am not sure Digia is making a lot of money after Qt was re-released under LGPL. There is some who will still buy a proprietary license due to fear of LGPL and there is some consulting involved but the market is certainly reduced than before. There is a significant amount of employees who have already been let go from Nokia. Also, your example of a dominant Apache isn't a good one anymore. Nginx has taken a significant share of the market.
"What more do you want than popular, successful, high quality, well documented, LGPL and with open governance?"
To recap, open governance that is true in more than in marketing managed by a non-profit. True multi-vendor participation instead of a dominant single vendor. No requirement of contributor license agreements. At a technical level, reduction and possibly eliminating the practise of duplicating existing system libraries with a Qt flavor. These would be a good start.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 18:04 UTC (Thu) by boudewijn (subscriber, #14185) [Link]
"To recap, open governance that is true in more than in marketing managed by a non-profit. True multi-vendor participation instead of a dominant single vendor. No requirement of contributor license agreements. At a technical level, reduction and possibly eliminating the practise of duplicating existing system libraries with a Qt flavor. These would be a good start."
No, it wouldn't be a good start, not for you. You would still be clamouring for something more, something else, something that's not in place yet -- you would invent another reason why you cannot use Qt, because you're too partisan towards GTK to allow for anything else.
Because the open governance is more than marketing. Because there are multipe vendors participating -- like kdab, for instance, or kde. Because there are sound technical arguments for not using the stl. Which you would never admit to, because your position is bound up with having to reject those arguments. Basically the only thing in your list that makes a modicum of sense as an argument for not contributing to Qt is the CLA -- and that doesn't make any sense at all as an argument for not using Qt.
But hey, I know I wasted my time here. It's vanishingly unlikely that any argument will change the mind of someone who is as entrenched in their position as you are.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 18:13 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]
"you would invent another reason why you cannot use Qt, because you're too partisan towards GTK to allow for anything else."
https://en.wikipedia.org/wiki/Straw_man
I never said I cannot use Qt. You just made that up.
Dricot: A freasy future for GNOME Posted Sep 6, 2012 13:30 UTC (Thu) by jospoortvliet (subscriber, #33164) [Link]
Sure GTK+ developers would contest anyone saying Qt is lightyears ahead but I think there are few people who can honestly argue it is not the case - of course, GTK+ has a much smaller scope than Qt but even then, it isn't very realistic to expect the handful of folks working on GTK to have kept up with the hundreds of engineers who were hacking on Qt for the last 10 years...
Maybe the splitting up of Qt in various libraries will help in the regard of the "do it our way" but it'll always be like that to some extend I suppose.
|
|||||||||||||