Hi, all,
i would like to hear some opinions about what
to use for GUI development with Lisp under Linux.
(I know there is comp.lang.lisp.x, but there is hardly any traffic)
Thank you in advance,
Matthias
Matthias Hanitzsch <··················@dynamic.de> writes:
> Hi, all,
>
> i would like to hear some opinions about what
> to use for GUI development with Lisp under Linux.
Well, I make fairly light use of gui toolkits (about the only
non-systems-level stuff that gets me excited are development tools),
but I use Garnet quite happily. I don't think a whole lot of people
use it anymore, but there is still an active user base, and the code
is mature enough, that it doesn't need a lot of maintenance. CLX is
low-level; it's just Lisp bindings to X11. While they're apparently a
lot more usable from Lisp than from C, you probably don't want to
write an application using them directly.
It might be helpful to know what you want to do, or what you tend to
do with your GUI toolkit. I'll let others chime in with their
preferred kits.
If you already know Motif or GTK, you might want to check out CLM and
CLG [clg is bindings to the current development branch of gtk, but
this is scheduled to become the stable branch Real Soon Now, so it
might be a reasonable option -- oh yeah, they do have bindings to
stable gtk, too, but the approach they're taking is changing
completely (and for the better) with the new branch].
You might also want to look at the CLiki page for Graphics Toolkits:
<http://ww.telent.net/cliki/Graphics%20Toolkit>
> (I know there is comp.lang.lisp.x, but there is hardly any traffic)
Eh, it's perfectly appropriate here. When Lisp takes over the world
alla C++, this should probably be just a language forum, but until
then, it's a broader Lisp forum.
--
/|_ .-----------------------.
,' .\ / | No to Imperialist war |
,--' _,' | Wage class war! |
/ / `-----------------------'
( -. |
| ) |
(`-. '--.)
`. )----'
···@conquest.OCF.Berkeley.EDU (Thomas F. Burdick) writes:
> > (I know there is comp.lang.lisp.x, but there is hardly any traffic)
>
> Eh, it's perfectly appropriate here. When Lisp takes over the world
> alla C++, this should probably be just a language forum, but until
> then, it's a broader Lisp forum.
Furthermore comp.lang.lisp.x _isn't_ about Lisp+X, it is about XLISP,
a completely different dialect of Lisp than Common Lisp (which is
mostly the only Lisp discussed in c.l.l nowadays). So the traffic
that isn't there is about XLISP, not Lisp+X ;)
Regs, Pierre.
--
Pierre R. Mai <····@acm.org> http://www.pmsf.de/pmai/
The most likely way for the world to be destroyed, most experts agree,
is by accident. That's where we come in; we're computer professionals.
We cause accidents. -- Nathaniel Borenstein
In article <··············@orion.bln.pmsf.de>, "Pierre R. Mai"
<····@acm.org> wrote:
> Common Lisp (which is mostly the only Lisp discussed in c.l.l
> nowadays)
It's not easy to talk about other lisps here because there are some
people who appear to think that comp.lang.lisp *means* Common Lisp.
I think this group should welcome compare-and-contrast and similar
discussions relating to lisp and other more or less closely-related
languages, with the language-specific stuff in the group dedicated to
that particular dialect (comp.lang.scheme, comp.lang.dylan,
comp.lang.lisp.x, comp.lang.lisp.mcl etc etc).
Unfortunately, it appears that Common Lisp is the *only* lisp that
doesn't have a dedicated newsgroup. (Well, there's emacs lisp, but
comp.emacs serves). I don't know why that is.
-- Bruce
Bruce Hoult wrote:
> In article <··············@orion.bln.pmsf.de>, "Pierre R. Mai"
> <····@acm.org> wrote:
>
>> Common Lisp (which is mostly the only Lisp discussed in c.l.l
>> nowadays)
>
> It's not easy to talk about other lisps here because there are some
> people who appear to think that comp.lang.lisp *means* Common Lisp.
>
> I think this group should welcome compare-and-contrast and similar
> discussions relating to lisp and other more or less closely-related
> languages, with the language-specific stuff in the group dedicated to
> that particular dialect (comp.lang.scheme, comp.lang.dylan,
> comp.lang.lisp.x, comp.lang.lisp.mcl etc etc).
Well - as you already list the newsgroups related to that "lisps" I think
it is most of the time a better idea to use this groups talk talk about
them.
I think it is ok if a topic implies comparison with other dialects or
languages - but then it is IMHO ok too, to mention how langages like C, ML
or whatever compare to lisp in that situation. It is a good thing to learn
from others.
> Unfortunately, it appears that Common Lisp is the *only* lisp that
> doesn't have a dedicated newsgroup. (Well, there's emacs lisp, but
> comp.emacs serves). I don't know why that is.
I think comp.lang.lisp _is_ the dedicated newsgroup for Common Lisp - it
seems to be at least for a significant amount of the readership of the
group. Maybe there is the wish for some people from other comunities (like
Dylan, Scheme...) to have a central battlefield where they could carry out
there general advocacy discussions but I do not feel such a need.
Btw.: Is the above mentioned the reason you answer every second question
here with "Dylan"? It sometimes seems to me there is more Dylanspecific
traffic here in c.l.l than in c.l.dylan ;-)
ciao,
Jochen
--
http://www.dataheaven.de
In article <············@rznews2.rrze.uni-erlangen.de>, Jochen Schmidt
<···@dataheaven.de> wrote:
> I think it is ok if a topic implies comparison with other dialects or
> languages - but then it is IMHO ok too, to mention how langages like C,
> ML or whatever compare to lisp in that situation. It is a good thing to
> learn from others.
I agree.
> > Unfortunately, it appears that Common Lisp is the *only* lisp that
> > doesn't have a dedicated newsgroup. (Well, there's emacs lisp, but
> > comp.emacs serves). I don't know why that is.
>
> I think comp.lang.lisp _is_ the dedicated newsgroup for Common Lisp
> - it seems to be at least for a significant amount of the readership
> of the group.
I agree that a certain proportion of the readership appears to think
that way.
But what does the group charter say? Does the group even have a
charter, or does it simply date from the depths of usenet time?
> Btw.: Is the above mentioned the reason you answer every second
> question here with "Dylan"?
Well it could be worse -- my point of reference could be VB or C++ or
something... :-)
Unfortunately I'm just a beginner in CL, though I'm trying to learn more
about it. Like many others, my first exposure to "lisp" was in a
university AI course (18 years ago) which definitely gave the impression
that Lisp 1.5 was the pinnacle of lisp development. So now I've ended
up coming back to lisp via the route of Dylan and that is my anchor
point in this field of knowledge, so that is what I compare other things
to -- especially since I know Dylan very well.
And so I often find myself in the position of trying to find out exactly
how CL and Dylan differ or are similar in some particular respect. Not
just on the surface (I hope) but the deep meanings, and the logic or
philosophy or historical accident behind the design decisions.
I hope that others don't mind too much, and perhaps the learning and
understanding can sometimes flow in both directions.
> It sometimes seems to me there is more Dylan specific traffic here
> in c.l.l than in c.l.dylan ;-)
At a conference I went to a couple of years ago, one speaker addressed
the question of why Linux was "beating" BSD in mindshare. His thesis
was that it was because the development of Linux was conducted very
publicly, warts and all, in newsgroups where any casual observer could
easily see that there was something happening. BSD development, on the
other hand, took place mostly on mailing lists and people had to expend
considerable effort (in comparison) to find and join the lists and read
the archives.
Over the last several years there has usually been a lot more going on
in, say, the Gwydion Dylan mailing lists than in c.l.dylan. Your
observation makes me wonder if perhaps the Linux vs BSD theory is
correct.
-- Bruce
Matthias Hanitzsch <··················@dynamic.de> writes:
> Hi, all,
>
> i would like to hear some opinions about what
> to use for GUI development with Lisp under Linux.
>
> (I know there is comp.lang.lisp.x, but there is hardly any traffic)
>
> Thank you in advance,
>
> Matthias
The main question is if you want to pay money for your development
project. If this is the case, I think the commercial offerings from
Xanalys and Franz can't be beaten as far as GUI development is
concerned. As you'll know, both offer free trial versions so you can
see for yourself if they suit your needs. I've played with the
LispWorks CAPI and was very impressed by it. I'm sure that Allegro's
Common Graphics (if that's the correct name) is also very good.
If you're looking for 'free' stuff: I've tried some of the toolkits
mentioned at <http://ww.telent.net/cliki/Graphics Toolkit> and I have
to admit that I didn't really fell in love with one of them. Most of
them aren't complete or aren't really well integrated into the Lisp
development systems 'cause they are ports of systems that were
originally intended for other language bindings. What I liked most was
Garnet because it seemed to be mature, efficient, and complete. The
drawbacks are that it uses its own object system instead of CLOS (I
guess this is due to historical reasons) and that it looks rather
old-fashioned. Also, it isn't maintained anymore expect for Fred
Gilham incorporating patches in his spare time.
To make this clear: I really admire the people who work or worked on
these open source projects and I thank them for their hard work. It's
just that the whole GUI area seems to be tedious and complicated and
not well-suited for uncompensated open source projects. But I think
this is not really a Lisp problem and thus I'll shut up here.
Just my two cents,
Edi.
PS: Another important question is whether you're planning to deliver
your application and to whom. This will sure be easier with, say,
LispWorks where you can just give an executable to your
customer. (Same with Allegro.) If, on the other hand, you want to
deliver on CMUCL or CLISP with one of the open source toolkits, you
have to ask your customers to install the Lisp environment _and_ the
GUI toolkit, too. Try this yourself and you'll see that it's not as
easy as GNU's "./configure && make && make install" stuff in many
cases.
···@agharta.de (Dr. Edmund Weitz) writes:
> Matthias Hanitzsch <··················@dynamic.de> writes:
>
> > Hi, all,
> >
> > i would like to hear some opinions about what
> > to use for GUI development with Lisp under Linux.
> >
> > (I know there is comp.lang.lisp.x, but there is hardly any traffic)
> >
> > Thank you in advance,
> >
> > Matthias
>
> The main question is if you want to pay money for your development
> project. If this is the case, I think the commercial offerings from
> Xanalys and Franz can't be beaten as far as GUI development is
> concerned. As you'll know, both offer free trial versions so you can
> see for yourself if they suit your needs. I've played with the
> LispWorks CAPI and was very impressed by it. I'm sure that Allegro's
> Common Graphics (if that's the correct name) is also very good.
>
> If you're looking for 'free' stuff: I've tried some of the toolkits
> mentioned at <http://ww.telent.net/cliki/Graphics Toolkit> and I have
> to admit that I didn't really fell in love with one of them. Most of
> them aren't complete or aren't really well integrated into the Lisp
> development systems 'cause they are ports of systems that were
> originally intended for other language bindings. What I liked most was
> Garnet because it seemed to be mature, efficient, and complete. The
> drawbacks are that it uses its own object system instead of CLOS (I
> guess this is due to historical reasons) and that it looks rather
> old-fashioned.
You do know that Garnet has two looks-and-feels, right? It defaults
to a clean, slim, classical look[*], but it also can look Motif-like.
Of course, Motif looks mid-90s, which isn't quite modern anymore, but
I don't think it's dated, yet.
> Also, it isn't maintained anymore expect for Fred Gilham
> incorporating patches in his spare time.
I assume this will eventually become a problem, but I think it's
mature enough that it doesn't really need much maintenance. Probably,
it'll survive as-is until Motif-alikes look dated, at which point
it'll fade away.
> To make this clear: I really admire the people who work or worked on
> these open source projects and I thank them for their hard work. It's
> just that the whole GUI area seems to be tedious and complicated and
> not well-suited for uncompensated open source projects. But I think
> this is not really a Lisp problem and thus I'll shut up here.
Very much agreed. Garnet's a gem (har har) because it's from CMU.
GUIs might not be well-suited to open-source development, but there
are some real successes. Lesstif, for one (it took forever, but it's
amazingly complete). GTK+, for another, actually. What with the
latest version, and with Glade, it's amazingly advanced. And it
strives to be language-neutral, as far as is practical. Hell, they've
even introduced C-language emulation of closures, explicitly for
implementors of bindings from dynamic languages. I'm honestly quite
impressed with the latest GTK+, and I think it will really be a
viable, multi-language, multi-platform toolkit into the future.
> PS: Another important question is whether you're planning to deliver
> your application and to whom. This will sure be easier with, say,
> LispWorks where you can just give an executable to your
> customer. (Same with Allegro.) If, on the other hand, you want to
> deliver on CMUCL or CLISP with one of the open source toolkits, you
> have to ask your customers to install the Lisp environment _and_ the
> GUI toolkit, too. Try this yourself and you'll see that it's not as
> easy as GNU's "./configure && make && make install" stuff in many
> cases.
Of course, you could distribute the FASL files yourself, or even a
dumped image.
[*] Oy, in my old age of 23, the vast majority of modern
looks-and-feels just hurt my eyes. They're so busy and just plain
... ugly. I love the older Mac L&F, pre-OpenStep NeXTStep, and stuff
like Garnet, and classical SmallTalk (eg, Squeak, or the old ST-80).
Maybe it's because I'm a bit of a typesetting dork, and I hate to see
half-assed attempts at looking fancy -- I'd rather see someone making
a smaller reach, and attaining it. Or maybe it's my astigmatism that
keeps worsening. I do know that I'm one of the few who *greatly*
prefers the look-and-feel of Emacs 20 over 21.
--
/|_ .-----------------------.
,' .\ / | No to Imperialist war |
,--' _,' | Wage class war! |
/ / `-----------------------'
( -. |
| ) |
(`-. '--.)
`. )----'
···@conquest.OCF.Berkeley.EDU (Thomas F. Burdick) writes:
> You do know that Garnet has two looks-and-feels, right? It defaults
> to a clean, slim, classical look[*], but it also can look
> Motif-like. Of course, Motif looks mid-90s, which isn't quite
> modern anymore, but I don't think it's dated, yet.
Er, to be honest, I once read that Garnet has two look-and-feels, but
it looks as if I forgot about it and only played with the classical
one. The classical look wasn't a problem for me (see remark below) but
I mentioned it because it might be important if you have customers in
mind.
> Very much agreed. Garnet's a gem (har har) because it's from CMU.
> GUIs might not be well-suited to open-source development, but there
> are some real successes. Lesstif, for one (it took forever, but
> it's amazingly complete). GTK+, for another, actually. What with
> the latest version, and with Glade, it's amazingly advanced. And it
> strives to be language-neutral, as far as is practical. Hell,
> they've even introduced C-language emulation of closures, explicitly
> for implementors of bindings from dynamic languages. I'm honestly
> quite impressed with the latest GTK+, and I think it will really be
> a viable, multi-language, multi-platform toolkit into the future.
Sounds nice. I haven't looked at it for a long time, and I think I
should do it again. (If I only had more time...)
What about platforms? Last time I looked there was a Windows port but
it seemed to be stuck. I also don't know of a port to Mac OS 'Classic'
or the native Aqua GUI to Mac OS X.
> [*] Oy, in my old age of 23, the vast majority of modern
> looks-and-feels just hurt my eyes. They're so busy and just plain
> ... ugly. I love the older Mac L&F, pre-OpenStep NeXTStep, and
> stuff like Garnet, and classical SmallTalk (eg, Squeak, or the old
> ST-80). Maybe it's because I'm a bit of a typesetting dork, and I
> hate to see half-assed attempts at looking fancy -- I'd rather see
> someone making a smaller reach, and attaining it. Or maybe it's my
> astigmatism that keeps worsening. I do know that I'm one of the few
> who *greatly* prefers the look-and-feel of Emacs 20 over 21.
:)
I'm a bit older but it's the same with me. I loved the old Mac OS
GUI-wise before it got more and more bloated. One thing that I value
much more than a spiffy look is a consistent user experience. I have
yet to find something that comes close to, say, System 7.5 in this
regard. Mac OS X is a nightmare IMHO. Microsoft could force developers
to adher to a consistent user interface due to their sheer market
power but they're too busy violating their own interface guidelines
all the time (if they _have_ something like this).
Ooops, no mention of Lisp in this article? Guess I should stop ranting
now.
Bye,
Edi.
···@agharta.de (Dr. Edmund Weitz) writes:
> ···@conquest.OCF.Berkeley.EDU (Thomas F. Burdick) writes:
>
> > Very much agreed. Garnet's a gem (har har) because it's from CMU.
> > GUIs might not be well-suited to open-source development, but there
> > are some real successes. Lesstif, for one (it took forever, but
> > it's amazingly complete). GTK+, for another, actually. What with
> > the latest version, and with Glade, it's amazingly advanced. And it
> > strives to be language-neutral, as far as is practical. Hell,
> > they've even introduced C-language emulation of closures, explicitly
> > for implementors of bindings from dynamic languages. I'm honestly
> > quite impressed with the latest GTK+, and I think it will really be
> > a viable, multi-language, multi-platform toolkit into the future.
>
> Sounds nice. I haven't looked at it for a long time, and I think I
> should do it again. (If I only had more time...)
>
> What about platforms? Last time I looked there was a Windows port but
> it seemed to be stuck. I also don't know of a port to Mac OS 'Classic'
> or the native Aqua GUI to Mac OS X.
Last I looked, the Win32 port was continuing to improve. It's in the
main source tree, now, I think. Of course, Windows and Unix gets
something like 90-95% of desktops and workstations (sadly), but since
OS X is also a unix, it even works there (with the rootless X server
installed). I don't know of any plans to port it to Quartz
(DisplayPDF), and really, and such plans would be suspicious, and
Apple has promised to sue anyone into oblivion who makes things that
look like Aqua. But GTK's "classic Mac" theme is decent, so it would
probably at least not scare Mac users :-).
As for GTK as a multiplatform GUI toolkit for Lisp, I don't think CLG
runs on anything other than CMUCL (I know that CLISP support got
dropped for lack of MOP compliance). It could probably be ported to
ACL, LW, &/or MCL, though (I think).
> Mac OS X is a nightmare IMHO.
Aqua is horrifyingly busy, but at least it can get somewhere close to
pulling it off. I'm just hoping that someone will, in good Mac
tradition, come up with a shareware patch to give it a NeXT-like L&F.
--
/|_ .-----------------------.
,' .\ / | No to Imperialist war |
,--' _,' | Wage class war! |
/ / `-----------------------'
( -. |
| ) |
(`-. '--.)
`. )----'
···@agharta.de (Dr. Edmund Weitz) wrote:
>Matthias Hanitzsch <··················@dynamic.de> writes:
>
>> Hi, all,
>>
>> i would like to hear some opinions about what
>> to use for GUI development with Lisp under Linux.
>>
>> (I know there is comp.lang.lisp.x, but there is hardly any traffic)
>>
>> Thank you in advance,
>>
>> Matthias
>
>The main question is if you want to pay money for your development
>project. If this is the case, I think the commercial offerings from
>Xanalys and Franz can't be beaten as far as GUI development is
>concerned. As you'll know, both offer free trial versions so you can
>see for yourself if they suit your needs. I've played with the
>LispWorks CAPI and was very impressed by it. I'm sure that Allegro's
>Common Graphics (if that's the correct name) is also very good.
yes it is Common Graphics and is a good product. Remember CG is MS
Windows only. I know a port to unix is in the plans for many years...
--
www.algo.be
Logo programming : www.algo.be/logo.html
Francis Leboutte <··········@algo.be> writes:
> yes it is Common Graphics and is a good product. Remember CG is MS
> Windows only.
That I didn't know. I thought they had some sort of cross-platform GUI
like CAPI and it was just the GUI builder that was Windows-only.
So, if you want to build a GUI interface to your ACL application that
runs on all supported platforms, you have to use their CLIM?
Thanks,
Edi.
···@agharta.de (Dr. Edmund Weitz) wrote:
>Francis Leboutte <··········@algo.be> writes:
>
>> yes it is Common Graphics and is a good product. Remember CG is MS
>> Windows only.
>
>That I didn't know. I thought they had some sort of cross-platform GUI
>like CAPI and it was just the GUI builder that was Windows-only.
>
>So, if you want to build a GUI interface to your ACL application that
>runs on all supported platforms, you have to use their CLIM?
Yes. The same if you want to develop for Unix only. In the past, there
was also a Franz GUI toolkit for Unix only. Apparently it has
disappeared from their offer, see
http://www.franz.com/products/gui_tools/index.lhtml
>Thanks,
>Edi.
--
www.algo.be
Logo programming : www.algo.be/logo.html
···@agharta.de (Dr. Edmund Weitz) writes:
> If you're looking for 'free' stuff: I've tried some of the toolkits
> mentioned at <http://ww.telent.net/cliki/Graphics Toolkit> and I have
> to admit that I didn't really fell in love with one of them. Most of
> them aren't complete or aren't really well integrated into the Lisp
> development systems 'cause they are ports of systems that were
> originally intended for other language bindings. What I liked most was
> Garnet because it seemed to be mature, efficient, and complete. The
> drawbacks are that it uses its own object system instead of CLOS (I
> guess this is due to historical reasons) and that it looks rather
> old-fashioned. Also, it isn't maintained anymore expect for Fred
> Gilham incorporating patches in his spare time.
I've found CLM in the CMUCL/Motif variant to be quite usable out of
the box, if you only need support for CMUCL and X+Motif. With a
couple of utility macros/functions, it can be as easy to use as CAPI
for a number of tasks (modulo not having an interface builder).
Since it is mentioned so seldomly, I took the liberty of updating the
empty entry on CLM on <http://ww.telent.net/cliki/Graphics Toolkit> to
include some more information.
> PS: Another important question is whether you're planning to deliver
> your application and to whom. This will sure be easier with, say,
> LispWorks where you can just give an executable to your
> customer. (Same with Allegro.) If, on the other hand, you want to
> deliver on CMUCL or CLISP with one of the open source toolkits, you
> have to ask your customers to install the Lisp environment _and_ the
> GUI toolkit, too. Try this yourself and you'll see that it's not as
> easy as GNU's "./configure && make && make install" stuff in many
> cases.
That too isn't very problematic with CMUCL/Motif. You just need to
include the server binary along with the lisp executable and image
files, and together with a small wrapper-script, it runs out of the
box on any system with a properly installed Motif. And if that is a
problem, you can just package a statically-linked server binary
instead (either commercial Motif, if you have the proper licence, or
Lesstif).
Regs, Pierre.
--
Pierre R. Mai <····@acm.org> http://www.pmsf.de/pmai/
The most likely way for the world to be destroyed, most experts agree,
is by accident. That's where we come in; we're computer professionals.
We cause accidents. -- Nathaniel Borenstein
"Pierre R. Mai" <····@acm.org> writes:
> (either commercial Motif, if you have the proper licence, or Lesstif).
Or even Open Motif if you're on an "Open Source" OS:
<http://www.opengroup.org/openmotif/license/>
Ole
ROHNE Ole <·········@cern.ch> writes:
> "Pierre R. Mai" <····@acm.org> writes:
>
> > (either commercial Motif, if you have the proper licence, or Lesstif).
>
> Or even Open Motif if you're on an "Open Source" OS:
> <http://www.opengroup.org/openmotif/license/>
Indeed. Though sadly Open Motif doesn't come with support for Motif
1.x (unlike commercial Motif or Lesstif), creating needless
incompatabilities for dynamically linked executables...
Regs, Pierre.
--
Pierre R. Mai <····@acm.org> http://www.pmsf.de/pmai/
The most likely way for the world to be destroyed, most experts agree,
is by accident. That's where we come in; we're computer professionals.
We cause accidents. -- Nathaniel Borenstein
Thank you for all your advice so far.
Here is what i forgot to mention:
I have just started to learn Lisp. And this "GUI development" is just
for
my own use. So i am searching for free stuff and i dont want to give it
away or sell it.
Furthermore i dont need "fancy" features, just the normal stuff
(windows,
menus, radiobuttons, ...).
Ok, that was just additional.
Matthias