Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Posted Jul 4, 2013 20:33 UTC (Thu) by anselm (subscriber, #2796)In reply to: Off-topic desktop whinge, unrelated to Fedora per se by rahulsundaram
Parent article: Fedora 19 released
Which is obviously the reason why so many ISVs from »outside the US and some of Europe« have opted to base their proprietary applications on GNOME and Gtk instead. After all, the market is completely flooded with those …
Posted Jul 4, 2013 23:19 UTC (Thu)
by rahulsundaram (subscriber, #21946)
[Link] (20 responses)
Posted Jul 4, 2013 23:51 UTC (Thu)
by anselm (subscriber, #2796)
[Link] (19 responses)
It was Cyberax who claimed that ISVs found the commercial Qt license too expensive, and I was replying to that. I have no opinion regarding custom in-house applications, which are obviously under different constraints than software for external commercial distribution.
Anyway, my main point was that GNOME was started not exclusively because of KDE's licensing issues with Qt but also because of NIH and dislike of C++ on the part of the early GNOME developers. If Qt's licensing had been the only problem, it would clearly have been less work at the time to come up with a freely-licensed Qt clone to support KDE than to write most of a new X11 toolkit and a complete new desktop environment from scratch, and that would not have led to the same degree of fragmentation that we see now.
The fact that GNOME co-founder Miguel de Icaza later went on to develop Mono tells us that he doesn't seem to have a problem in principle with trying to clone huge pieces of very proprietary software such as .NET. Compared to the Microsoft stuff, Qt in the late 1990s was positively liberally licensed. Also, once a hypothetical free Qt clone had been finished to the degree that KDE ran on it, the KDE community could simply have abandoned the original Qt, thereby removing the necessity to track Qt forever the way Miguel de Icaza gets to track .NET.
Posted Jul 5, 2013 1:16 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (18 responses)
>Also, once a hypothetical free Qt clone had been finished to the degree that KDE ran on it, the KDE community could simply have abandoned the original Qt, thereby removing the necessity to track Qt forever the way Miguel de Icaza gets to track .NET.
Also, you might note that Mono right now is actually used on more machines than the total amount of classic Linux desktops. Mostly because it is THE ONLY real language and framework that can be used to create software for ALL of the major mobile devices.
Posted Jul 5, 2013 6:57 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (17 responses)
So you say. I suppose that if one has no legal qualms about cloning .NET, what would keep one from cloning Qt? After all, Microsoft probably has more lawyers in the East Podunk, North Dakota, branch office than Trolltech ever had employees in the whole company.
And what does that have to do with anything? Presumably if Miguel de Icaza had devoted the same amount of energy to cloning Qt that he later put into cloning .NET all those machines might now be running his free Qt clone.
Posted Jul 5, 2013 7:03 UTC (Fri)
by mjg59 (subscriber, #23239)
[Link] (1 responses)
Based on personal experience: it's less difficult to write something with equivalent functionality than it is to write a re-implementation of something with a specified API.
Posted Jul 5, 2013 9:17 UTC (Fri)
by anselm (subscriber, #2796)
[Link]
Probably. After all, everyone knows that a college student would never be able to come up with something really big and complicated – like, say, an industry-strength Unix-like operating system kernel.
Posted Jul 5, 2013 7:04 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (14 responses)
Any realistic API-compatible clone of QT would be deemed a derived work because of the overwhelming number of literal copying that'll be required.
> And what does that have to do with anything? Presumably if Miguel de Icaza had devoted the same amount of energy to cloning Qt that he later put into cloning .NET all those machines might now be running his free Qt clone.
Posted Jul 5, 2013 7:31 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (13 responses)
If there is stuff that you need to copy literally because it is the only way to do something then that stuff is probably not eligible for copyright in the first place. Given that copyright is supposed to protect the output of the creative process, generally a work needs to be creative in order to be copyrightable, and if there is only one possible way of writing something down then coming up with that way is not especially creative.
Posted Jul 5, 2013 7:53 UTC (Fri)
by dark (guest, #8483)
[Link] (12 responses)
Posted Jul 5, 2013 9:13 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (11 responses)
This is what clean-room techniques are for. You have someone write a spec and/or tests and have somebody else, who hasn't seen the original (Qt or whatever) source, write the actual implementation to suit the spec and tests.
You obviously can't copy something that you have never seen, and anyway your own implementation will probably differ enough in detail to no longer look like an outright copy of the original to begin with.
It is probably worth noting that at the time we're talking about (late 1990s), Qt used to be considerably smaller and simpler than it is now.
Posted Jul 5, 2013 18:14 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (10 responses)
And anyway, you have now moved on from stuff that a small company or a collection of individuals can do to the stuff that requires a large company with a legal department.
Now think about it, suppose that I spent $100^500 to clone QT - what do I achieve?
Posted Jul 5, 2013 21:03 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (9 responses)
Nobody needs to clone Qt now. For one, Qt is now available under a reasonable selection of licences, and secondly, Qt is by now uncomfortably large and complicated.
Cloning Qt instead of implementing GNOME and most of Gtk from scratch would have been a more reasonable proposition 15 years ago when Qt was a much smaller piece of software. We might have avoided the desktop environment fragmentation that plagues us now if we could have standardised on KDE at a fairly early stage instead of coming up with a competing desktop environment.
On the other hand, chances are that even if Qt had been available under the LGPL from the beginning, the GNOME people would still have started GNOME just to be contrary. GNOME was initially as much about NIH and not wanting to use C++ as it was about Qt licensing.
Posted Jul 5, 2013 21:09 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (8 responses)
What do you do? Spend millions to clone a crappy UI toolkit with a guaranteed lawsuit afterwards? And what for, just to appease a bunch of developers of a marginal desktop environment (KDE)?
Nope, GTK creation was an entirely logical step at that time. Dislike of C++ might have played a role, but it wasn't decisive.
Of course, NOW we have QT under a reasonable license and it's much better than GTK. But we're talking about ancient history.
Posted Jul 5, 2013 21:25 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (7 responses)
If it had been up to me personally, I would have put the same effort that the GNOME people put into making Gtk usable (it was really bad at the time) into Tk instead (which was freely available and a lot farther along than Gtk). But of course the GNOME people didn't like that idea, either – NIH, remember –, so they basically had to roll their own no matter what. A bit like what the Canonical folks are doing today.
Posted Jul 5, 2013 21:39 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (6 responses)
It had its own problems and idiosyncrasies: it was not themeable, it had poor cross-platform support, no support for TrueType fonts and so on. Oh, and it also was butt-ugly.
"Approaching user interface design with a programmer's mindset can only end in pain, tragedy, and Tk." (c) David Vallner.
Posted Jul 5, 2013 23:27 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (5 responses)
Tk has been themable for a number of years now; there is no reason to believe that with appropriate effort that could have been built in earlier. As far as cross-platform support goes, you are misinformed; in the late 1990s Tk ran on Windows and the Mac as well as on X11 (including unusual X11 platforms like VMS), with native-looking widgets. In the late 1990s, TrueType font support was basically up to the X server, so if your X server supported them then Tk would too. (Now that we have XRENDER Tk of course supports that.) »Ugly« is in the eye of the beholder; Tk was originally built to look like Motif, which in the early 1990s was the obvious thing to emulate. In the meantime tastes have changed, and with Tk's theme support there is no reason why a Tk GUI should look any different than a Qt or Gtk GUI.
Tk also had some good things going for it in the late 1990s, such as very flexible geometry management, reasonable UTF-8 support (which at the time no other popular toolkit bothered with), or the canvas and text widgets which were really quite useful. It always seemed silly to me that somebody would spend vast amounts of effort to make Gtk into something that anyone except the GIMP developers would actually want to use; the same amount of work put into Tk would easily have taken care of all the things that people like bashing Tk for and then some. But then of course there is no accounting for taste, and »NIH« is a very powerful motivator.
Posted Jul 6, 2013 1:03 UTC (Sat)
by Cyberax (✭ supporter ✭, #52523)
[Link] (4 responses)
Sorry, you can't have it both ways. Either GTK was a logical step to create a free UI toolkit, or KDE caused huge fragmentation by choosing a proprietary UI toolkit instead of working with free UI toolkit developers.
Posted Jul 6, 2013 12:27 UTC (Sat)
by anselm (subscriber, #2796)
[Link] (3 responses)
When KDE started out there were no other free desktop environments (there were lots of free applications but nothing that tried to present a unified »look and feel« similar in scope to, say, that of the Mac). KDE took its inspiration from CDE, which at the time wasn't free at all. You can't really cause fragmentation if you're the first on the scene.
Personally I would of course have preferred to see a free desktop environment based on something like Tk rather than Qt. However, the KDE developers did not see fit to consult me before they got started ;^) You would have to ask them (more specifically, Matthias Ettrich) to find out exactly why they picked Qt. There is some information about their reasoning available on the KDE web site, but that very likely includes some hindsight.
Posted Jul 6, 2013 19:20 UTC (Sat)
by Cyberax (✭ supporter ✭, #52523)
[Link] (2 responses)
GTK developers faced the same exact situation, except that they were against proprietary Qt.
Posted Jul 6, 2013 19:33 UTC (Sat)
by anselm (subscriber, #2796)
[Link] (1 responses)
Your comment only tells us that you apparently haven't done much with Tk.
The GNOME people took an atrocious toolkit (Gtk) that had very little traction at the time and was barely good enough to support one single application (The GIMP) and spent massive effort to make it into something they could use for their written-from-scratch desktop environment. They could have put the same sort of effort into Tk and made it into something vastly better with less work. Tk, for its flaws, is really quite well-engineered and well-documented and it would have been absolutely feasible at the time to build a desktop environment based on it (at least as much so as building one based on what was then Gtk). I'd personally be inclined to think the main reason the GNOME people didn't do that is NIH and/or ignorance and/or not wanting to work with the Tk community.
Posted Jul 6, 2013 20:59 UTC (Sat)
by mpr22 (subscriber, #60784)
[Link]
Posted Jul 5, 2013 1:11 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link] (2 responses)
BTW, actually try look at commercial Unix software - I think most of it used GTK back then.
Oh, and if we're talking about fragmentation - why hasn't KDE used Motif? After all, it's totally free with JUST a $5000 per developer pricetag! I totally can't imagine why people would decide to write something that doesn't require forking over tons of bucks, with a vendor lock-in.
Posted Jul 5, 2013 7:11 UTC (Fri)
by anselm (subscriber, #2796)
[Link] (1 responses)
However bad Qt may have been at the time, Gtk was many times worse. Its main raison d'être was supporting The GIMP, and it took a major effort on the part of the early GNOME crowd to get it anywhere near where Qt was. Which, looking back, seems all the more silly given that there were better free alternatives around – and supports the thesis that GNOME was at least as much about NIH than it was about Qt licensing.
That which didn't use Motif or even other things like Tk which were more popular. (Tcl/Tk used to be pretty big in the (commercial) VLSI design scene, for example.) AFAIR, Gtk at the time had a fairly small slice of the cake as far as commercial development went.
Posted Jul 5, 2013 7:16 UTC (Fri)
by Cyberax (✭ supporter ✭, #52523)
[Link]
>Which, looking back, seems all the more silly given that there were better free alternatives around – and supports the thesis that GNOME was at least as much about NIH than it was about Qt licensing.
> That which didn't use Motif or even other things like Tk which were more popular. (Tcl/Tk used to be pretty big in the (commercial) VLSI design scene, for example.) AFAIR, Gtk at the time had a fairly small slice of the cake as far as commercial development went.
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Sure, and the Boston bomber is simply an angel, compared to Hitler. Why people want to keep him imprisoned?
Nope. Cloning was never a solution, from the legal standpoint.
Off-topic desktop whinge, unrelated to Fedora per se
Cloning was never a solution, from the legal standpoint.
Also, you might note that Mono right now is actually used on more machines than the total amount of classic Linux desktops.
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
It turns out that it's far easier to re-implement an existing API when you have a large workforce funded by the sale of a business than it is when you're a college student.
Off-topic desktop whinge, unrelated to Fedora per se
Microsoft explicitly allowed cloning, by standardizing .NET. They've also granted a royalty-free license for their patents on core .NET technologies.
See above.
Off-topic desktop whinge, unrelated to Fedora per se
Any realistic API-compatible clone of QT would be deemed a derived work because of the overwhelming number of literal copying that'll be required.
But Qt is C++, which means a lot of the implementation can be in the interface. For example qlist.h contains the entire QList implementation (mostly as templates). On the one hand, it would take an expert to figure out what you can change without breaking compatibility. On the other hand, it's 900 lines of complicated code and it seems unlikely that a court will decide there's no creative expression in there.
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Now think about it, suppose that I spent $100^500 to clone QT - what do I achieve?
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Nope, GTK creation was an entirely logical step at that time. Dislike of C++ might have played a role, but it wasn't decisive.
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Maybe they perceived Tk as strongly coupled to Tcl, and didn't like Tcl.
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Off-topic desktop whinge, unrelated to Fedora per se
Most ISVs simply looked at Linux and decided it's not worth it to fork over tons of dollars just for a crappy UI toolkit (and QT _was_ a crappy UI toolkit back then).
BTW, actually try look at commercial Unix software - I think most of it used GTK back then.
Off-topic desktop whinge, unrelated to Fedora per se
Duh. So what? It was obvious that some time will be required to reach a level where GTK is better than QT for most applications.
Which alternatives? Lesstif? Or maybe early EFL?
VMWare used it, several commercial suite also used it. Then lots of software started to use Eclipse which on Linux is based on GTK.