Joachim Durchholz wrote:
>> For GUIs that don't completely suck, Linux only has GTK+ (written in C
>> with reference counting) and Qt (a commercial product tightly bound to
>> C++).
>
> I think there are FPL equivalents to GTK+, but I don't know their
> status. It's also a bit difficult to tell which of them will continue to
> be maintained.
None have anything like the traction of GTK+ or Qt though. There are also
projects like FLTK.
>> The nearest thing we have to a CLR is the JVM, which is another product
>> of commerce. But the JVM has basically no decent languages with any
>> sup****t. The nearest is probably Scala which is quite a sucky language
>> compared to ML and has 40x fewer users than OCaml.
>
> Dunno, Scala have been described as interesting and useful here.
> Personally, I found the design goals extremely interesting but don't
> know how they have been working out in practice.
>
> The JVM isn't a good target for FPLs though. To eliminate tail calls you
> need to pack everything into a single function, but then functions are
> limited in size.
> Packing everything into a single function also requires additional
> consideration when providing APIs for outside code.
> Both effects seem to be quite effective deterrents for FPL implementors.
>
> The JVM isn't very flexible for other languages, too. Eiffel compilers
> tend to generate single-class programs. I wouldn't want to ****t C or C++
> to the JVM either. Whatever the "****t X to the JVM" project was, all
> re****ted considerable problems when trying to make the generated code
> interoperate with Java.
> So the general consensus seems to be that the JVM is a mostly-Java
affair.
Yes. Hopefully they'll fix this. They have made noises about it, e.g.
providing proper tail calls.
>> In reality, my Linux box only recently acquired Flash in the browser
and
>> that is only if I use a different web browser (Firefox rather than
>> Konqueror).
>
> But you *can* replace Konqueror with Firefox.
No, you can't. Konqueror is integrated into KDE so Firefox can only
replace
the web browser aspect of Konqueror.
> The base problem being that Flash is proprietary, and too complex to
> reverse engineer in an afternoon. (Probably more complex than necessary,
> too.)
I was under the impression that Flash was now open and could be
implemented
very easily if you know what you're doing. Indeed, several people have
asked me to do exactly that: write a Flash front-end for Smoke.
>> You can completely ignore whatever Microsoft say. Just look at what
>> they've built with .NET: Libraries as capable as Java's. Languages as
>> capable as OCaml. Performance as good as C++ for numerics. There is
some
>> real sophistication in the CLR. F# has lots of useful features like
>> active patterns that the open source community never created.
>
> Yes, and lock-in mechanisms as well...
Sure. That's just business as usual, of course.
> ... which is why I have stayed away and probably will (unless somebody
> is paying me well to work on or for .net), and, of course, the reason
> why I don't really know how much of these positive characteristics hold
> water.
> I once had seen a glimmer of GDI+ (wich, I understand, is the .net
> equivalent of GDI). I noticed that the design was noticeably better than
> that of GDI, but also found that some mistakes still were being
repeated.
The WPF API is hideous and its performance is an order of magnitude worse
than Smoke's but it is still far better than anything available for Linux.
> On the other hand, quite a few *very* innovative things were done.
> Just one example: since the 2.6 kernels, Linux can handle thousands of
> processes without running out of kernel objects or putting on much of an
> overhead; on Windows or pretty much every other Unixoid OS, you're still
> stuck with threading if you want performance and parallel processing.
That underpins OCaml's concurrency so I'm all for that. ;-)
>> and botched a rather crummy Windows look and feel on top.
>
> Yes, but you're ba****ng the wrong community here.
> The Linux folks are quite different from the X folks, which are still
> quite different from the Gnome and KDE folks.
> If you wanted to innovate over Microsoft's GUI stuff, you'd need a
> competitor for Gnome/KDE (and possibly one to X).
Well, ok. True. I'm thinking in terms of competing on the desktop, of
course. Whatever is available for Linux there is essentially an accident
anyway, so there is little point in comparing "designs"...
> > There's very little innovation there compared to what MS are doing
> > now.
>
> I think it's less that MS is innovating so much, it's more that the GUI
> folks in the FOSS camp aren't doing well enough. I suspect too many
> people simply underestimate what should go into a well-designed GUI.
I agree entirely.
> > I think that's a good thing though: its about time MS had to earn a
>> living.
>
> Hm. MS has been earning a living for years now, with billions of cash
> and short-term money on its hands.
> It's about time they stop sucking more money than they have been worth
> out of the programming community.
That's exactly what I meant, yes.
--
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/products/?u
|
