On Fri, 07 Feb 2014 04:40:40 -0600
Tim Holy <
tim....@gmail.com> wrote:
Hi Tim,
> Gour, if you'd like to start a Qt.jl package, that would be great. If
> you succeeded in overcoming the major hurdles, I suspect there would
> be considerable interest. (The reason I started Cpp.jl was in part
> aiming towards Qt.)
I'm glad there are people here considering that Qt bindings might be
good thing for Julia.
Let me say tyhat yesterday evening I spent some time reading Julia's
docs on my small 7" tablet and I must say I'm *very* impressed with the
Julia as a language. It looks there is finally elegant language allowing
to use FP paradigm (amongst others), no need to take care about pointers
& other low(er)-level stuff, clean syntax, lots of library support, easy
C calling (FFI), great performance...
In short: Julia seems to be 'fine general programming language'. :-)
Based on everything above, I strongly believe that Qt bindings would be
killer feature!!
> Given that Python has wrapped Qt, I agree it's surely possible for
> Julia as well. But keep in mind that Python has been around a long
> time, has gone through several iterations of Qt wrappers, and
> collectively those wrappers have had probably tens-of-thousands of
> human-hours thrown at them (including corporate-sponsored hours).
I'm aware of that, but let's try to focus on 'is possible'. PyQt is nice
example of having decent high(er)-level bindings for C++ toolkit.
> In contrast, Gtk.jl is basically the work of 4 people (of whom Jameson
> is by far the most significant), and it's extraordinary how _pretty_
> the guts of it are and how productive that effort has been. Gtk.jl is
> still evolving, but it's clearly headed in a great direction.
Well, the GTK+ team itself is not very crowded place as well.
> I would choose the native Gtk.jl over a python/Qt intermediary
> without question---I did some performance profiling a few months
> back, and going through python was not pretty. I'm talking about
> factors of 10-100 here, and as a "big data" guy who wants to plot
> large numbers of points, in my book that performance gap utterly
> rules out python as a long-term foundation for graphics in Julia.
Do you mean PyCall.jl or straight PyQt/Python?
Iirc, I stumbled upon some benchmarks recently where Julia was compared
with Python/Cython and the difference was no so great, but not sure
about the details?
I'm also looking for a decent language for 'general desktop' application
involving some number crunching (e.g. we need to use 3rd party C library
for calculating planetary ephmeris), but we need good GUI on top of
that.
> With regards to whether Tobi is taking a big risk developing the IDE
> in Gtk.jl: my experience in porting ImageView from Tk.jl to Gtk.jl
> was that doing the ImageView port was only about 10% of the work
> (i.e., not too bad), 90% was improving Gtk.jl itself. In other words,
> even if a Qt.jl package eventually appears, switching Julietta over
> to it would be vastly less effort than developing Qt.jl in the first
> place.
In regard to Tobi's work, intention of my question was not solely
Julietta itself, but Qt bindings in general and e.g. why not PyCall with
Qt?
My interest is not so much in which GUI is IDE for the language written
'cause I believe that vim can serve me for quite some time on the way
while working on the app.
> I think it's possible to be of two minds: if you search old archives
> you'll see I had my own reservations about choosing Gtk+ rather than
> sucking it up and wrapping Qt. But having dived into Gtk, I'd say
> it's pretty darn good (at least on Linux, I haven't tested on other
> platforms).
I'm using Linux since '99 and only for short time I was using KDE -
sometime in the 0.9 era iirc. The rest of the time it was either GNOME2
or mostly XFCE and now, since some time I use plain i3.
So I always liked/preferred GTK, but trying to be pragmatical, I do not
believe there is much future for GTK except on Linux/GNOME.
Qt is simply superior in so many aspects today, starting with the docs,
number of devs/community, OS support etc.
However, as someone who *might* be just interested to *use* Julia and
its ecosystem, but probably without time to work (I can afford to only
work on my hobby open-source project in my free time) on the project of
the Qt.jl-size, I will stop here.
Sincerely,
Gour
--
The working senses are superior to dull matter; mind is higher
than the senses; intelligence is still higher than the mind;
and he [the soul] is even higher than the intelligence.