Harlequin Dylan on Unix 
Author Message
 Harlequin Dylan on Unix

Are there any Unix ports of Harlequin Dylan in the works?   It would be
nice if, at least, non-graphical programs could be run in a Unix
environment.  That way if you had server-side Dylan programs and ran
into scalability programs you could move things to a bigger box.  I
don't see any need for the GUI stuff and IDE on Unix, but to have the
command line tools and non-graphical libraries available on several
platforms would be a big plus.


Fri, 14 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:
> Are there any Unix ports of Harlequin Dylan in the works?  It would
> be nice if, at least, non-graphical programs could be run in a Unix
> environment.  That way if you had server-side Dylan programs and ran
> into scalability programs you could move things to a bigger box.  I
> don't see any need for the GUI stuff and IDE on Unix, but to have
> the command line tools and non-graphical libraries available on
> several platforms would be a big plus.

Being careful to avoid promising unannounced software, I will say:

 - our web site is open about our desire to port Dylan to other
   platforms, we would love to hear opinions as to which platforms are
   the most important. Please fill in the questionnaire found when
   downloading our Personal Edition (I don't have the address to hand,
   it has been posted here before), note that you can fill it in
   without having to actually download anything.

 - both of our other language products (LispWorks and MLWorks) work on
   a variety of Unix platforms as well as Windows 95/98 and NT, and
   most of Harlequin's Dylan developers come from a Unix
   background. We actually bootstrapped Dylan initially using a Dylan
   emulator in LispWorks on Unix.

 - we worked very hard with our Dylan partners to create standard
   library APIs that would be portable to multiple platforms. Gwydion
   Dylan is a concrete example that this effort succeeded, as to a
   lesser extent is the Apple Dylan Technology Release.

 - DUIM, our GUI programming toolkit, is one such library, and the
   Gwydion group plan to port it to the GTK toolkit sometime
   relatively soon (probably once Motley and the C-FFI are
   implemented). The authors of DUIM (Scott McKay and myself) both
   have a great deal of experience writing portable GUI toolkits in
   Lisp, and we are certain that DUIM will port extremely easily to
   all modern GUI toolkits.

 - our compiler is written entirely in (Harlequin) Dylan, using the
   standard libraries to read files, check timestamps etc. The only
   part of the compiler that needs to be written for a new platform
   is the native code generator.

 - our development environment is almost entirely written in Dylan, with
   all of the user interface implemented in DUIM (which as I've
   mentioned before is a portable toolkit). The only part that is
   implemented in C is our 'tether' technology for debugging
   applications both locally and remotely, but this too is wrapped up
   in a portable Dylan interface.

You are welcome to draw your own conclusions. :-)

In the meantime, we are working with the Gwydion group to standardize
a number of common libraries, so code written with the next release of
Harlequin Dylan will work directly in Gwydion Dylan on Unix (except
for our proprietary extensions for CORBA, COM, OLE etc).

We'd love to talk to you privately about your Harlequin Dylan needs,
if you'd be interested in doing so then please email

Cheers,

Andy

----------

Harlequin Inc.



Sat, 15 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix


Quote:

> > Are there any Unix ports of Harlequin Dylan in the works?  It would
> > be nice if, at least, non-graphical programs could be run in a Unix
> > environment.  That way if you had server-side Dylan programs and ran
> > into scalability programs you could move things to a bigger box.  I
> > don't see any need for the GUI stuff and IDE on Unix, but to have
> > the command line tools and non-graphical libraries available on
> > several platforms would be a big plus.

> Being careful to avoid promising unannounced software, I will say:

>  - our web site is open about our desire to port Dylan to other
>    platforms, we would love to hear opinions as to which platforms are
>    the most important.

Andy,

Any chance of Harlequin targeting Dylan to a universal platform like the
Java VM?
If so, Dylan could become available on a large number of CPUs/OSes. I
believe
the biggest contribution Java made to software development is not the Java
language, but the Java VM; which unfurtunately is under utilized by
programming
languages that compete with Java.

Regards,

Werner



Sat, 15 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix


    Date: Tue, 30 Mar 1999 22:34:40 +0200



    >
    > > Are there any Unix ports of Harlequin Dylan in the works?  It would
    > > be nice if, at least, non-graphical programs could be run in a Unix
    > > environment.  That way if you had server-side Dylan programs and ran
    > > into scalability programs you could move things to a bigger box.  I
    > > don't see any need for the GUI stuff and IDE on Unix, but to have
    > > the command line tools and non-graphical libraries available on
    > > several platforms would be a big plus.
    >
    > Being careful to avoid promising unannounced software, I will say:
    >
    >  - our web site is open about our desire to port Dylan to other
    >    platforms, we would love to hear opinions as to which platforms are
    >    the most important.

    Any chance of Harlequin targeting Dylan to a universal platform
    like the Java VM?  If so, Dylan could become available on a large
    number of CPUs/OSes. I believe the biggest contribution Java made
    to software development is not the Java language, but the Java VM;
    which unfurtunately is under utilized by programming languages
    that compete with Java.

I bet that some of the reasons the Java VM is under-utilized include:
 - It's not really very fast.  I hear that even Sun has Java VMs that
   run muc faster than the official Java VM.
 - It's not very good at supporting higher-order languages (languages
   with first-class functions).
 - It's got a built-in O-O model that directly supports only single
   dispatch in a single-inheritance class model.
 - It's got a built-in exception model that supports only exceptions
   that "unwind" before any handlers are called.
All of these problems can be worked around, but each one of these
things incurs a further performance penalty for languages like Dylan.



Sat, 15 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:


>     Date: Tue, 30 Mar 1999 22:34:40 +0200
>     Any chance of Harlequin targeting Dylan to a universal platform
>     like the Java VM?  If so, Dylan could become available on a large
>     number of CPUs/OSes.

      Maybe, maybe not.  All JVMs are not all created equal.  Granted they
      are more the same than most CPU/OS combinations but differences
      will likely arise.  

Quote:
>     to software development is not the Java language, but the Java VM;
>     which unfurtunately is under utilized by programming languages
>     that compete with Java.

    That's because it is the *JAVA* virtual machine.  Not the
    language neutral virtual machine.   Colusa had more language neutral
    technology, but they were assimilated by the Redmond collective.
    To run another language on the JVM means either conforming to Java's ways
    of doing things or essentially writing a virtual machine on top of a
    virtual machine.  The latter probably isn't going to go very fast even
    with just-in-time compilation.  That would just remove one layer. There
    are now two.

    It makes for a nice demonstration project. I'm not sure if it is a
    desirable approach for delivery.  Unless, the primary delivery target
    is the "web".  

Quote:
> I bet that some of the reasons the Java VM is under-utilized include:
>  - It's not really very fast.  I hear that even Sun has Java VMs that
>    run muc faster than the official Java VM.

    The "it will be ready 'real soon now'" Hot Spot. :-)

            http://java.sun.com/products/hotspot/index.html

--

Lyman



Sat, 15 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:


>     Date: Tue, 30 Mar 1999 22:34:40 +0200



>     > > Are there any Unix ports of Harlequin Dylan in the works?  It would
>     > > be nice if, at least, non-graphical programs could be run in a Unix
>     > > environment.  That way if you had server-side Dylan programs and ran
>     > > into scalability programs you could move things to a bigger box.  I
>     > > don't see any need for the GUI stuff and IDE on Unix, but to have
>     > > the command line tools and non-graphical libraries available on
>     > > several platforms would be a big plus.

>     > Being careful to avoid promising unannounced software, I will say:

>     >  - our web site is open about our desire to port Dylan to other
>     >    platforms, we would love to hear opinions as to which platforms are
>     >    the most important.

>     Any chance of Harlequin targeting Dylan to a universal platform
>     like the Java VM?  If so, Dylan could become available on a large
>     number of CPUs/OSes. I believe the biggest contribution Java made
>     to software development is not the Java language, but the Java VM;
>     which unfurtunately is under utilized by programming languages
>     that compete with Java.

Java uses the same vm paradigm as Smalltalk does.  Smalltalk vm is faster
than java's too.

- Show quoted text -

Quote:

> I bet that some of the reasons the Java VM is under-utilized include:
>  - It's not really very fast.  I hear that even Sun has Java VMs that
>    run muc faster than the official Java VM.
>  - It's not very good at supporting higher-order languages (languages
>    with first-class functions).
>  - It's got a built-in O-O model that directly supports only single
>    dispatch in a single-inheritance class model.
>  - It's got a built-in exception model that supports only exceptions
>    that "unwind" before any handlers are called.
> All of these problems can be worked around, but each one of these
> things incurs a further performance penalty for languages like Dylan.



Sun, 16 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:

> In the meantime, we are working with the Gwydion group to standardize
> a number of common libraries, so code written with the next release of
> Harlequin Dylan will work directly in Gwydion Dylan on Unix (except
> for our proprietary extensions for CORBA, COM, OLE etc).

I wouldn't consider CORBA proprietary. It's on our to-do list,
although not exactly on the top.

Andreas

--
"It is by caffeine alone I set my mind in motion.
It is by the Beans of Java that thoughts acquire speed,
the hands acquire shaking, the shaking becomes a warning.
It is by caffeine alone I set my mind in motion." -- National Lampoon's "Doon"



Mon, 17 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:
> > In the meantime, we are working with the Gwydion group to
> > standardize a number of common libraries, so code written with the
> > next release of Harlequin Dylan will work directly in Gwydion
> > Dylan on Unix (except for our proprietary extensions for CORBA,
> > COM, OLE etc).

> I wouldn't consider CORBA proprietary. It's on our to-do list,
> although not exactly on the top.

Sorry if that wasn't clear, I didn't mean that CORBA, COM, OLE are
proprietary to Harlequin, obviously anyone can add their own support
for any of them. I just meant that Harlequin provides a lot of support
for these technologies in the form of libraries, our own Dylan ORB,
and in our IDE, and we won't be sharing any of this with Gwydion. We
do have to leave ourselves some commercial possibilities so that we
can afford to do Dylan!

Andy

----------

Harlequin Inc.



Mon, 17 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:

> We do have to leave ourselves some commercial possibilities so that we
> can afford to do Dylan!

I think it's time for my traditional plug of Harlequin Dylan. ;-) Reasons
to buy:

 * Harlequin Dylan supports Windows (quite well, in fact).
 * Harlequin Dylan generates very efficient code.
 * Harlequin Dylan has an IDE.
 * Harlequin Dylan has a good de{*filter*}.
 * Harlequin Dylan supports incremental compilation.
 * The Enterprise Edition supports remote debugging.

Or to put it more briefly:

 * Dylan is a great language (but I may be partisan about this).
 * Harlequin has the best Dylan compiler.
 * Harlequin has the *only* Dylan compiler for Windows.
 * Professional Edition and Enterprise Edition are competitively priced
   in their respective market segments.
 * Harlequin is exceptionally supportive of open standards.

The reasons for using Gwydion Dylan are a bit different:

 * You have a sick and overwhelming compulsion to stay up until 3am working
   on Dylan tools which you'll give away.
 * You want to port Dylan to some operating system that Harlequin won't
   support in a million years. Oh, sure, they might port to the major
   flavors of Unix, but you're running OpenBSD for the StrongARM or BeOS on
   a PowerPC, and you'd rather spend a week porting a Dylan compiler than
   switch to another platform.
 * You've bought into this whole open source thing in a big way.
 * You want money, fame, power and influence. (Well, this *is* the first
   of April, right?)
 * You missed the glory years of Linux and always felt disappointed that
   you didn't have to write all your own device drivers.

In short, we serve very different markets. However, we're both committed to
Dylan standards. In the future, it should be possible to write non-trivial
tools under Gwydion Dylan and recompile them with Harlequin Dylan, or vice
versa.

Cheers,
Eric



Mon, 17 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:


>     Date: Tue, 30 Mar 1999 22:34:40 +0200

>     Any chance of Harlequin targeting Dylan to a universal platform
>     like the Java VM?  If so, Dylan could become available on a large
>     number of CPUs/OSes. I believe the biggest contribution Java made
>     to software development is not the Java language, but the Java VM;
>     which unfurtunately is under utilized by programming languages
>     that compete with Java.

> I bet that some of the reasons the Java VM is under-utilized include:
>  - It's not really very fast.  I hear that even Sun has Java VMs that
>    run muc faster than the official Java VM.
>  - It's not very good at supporting higher-order languages (languages
>    with first-class functions).
>  - It's got a built-in O-O model that directly supports only single
>    dispatch in a single-inheritance class model.
>  - It's got a built-in exception model that supports only exceptions
>    that "unwind" before any handlers are called.
> All of these problems can be worked around, but each one of these
> things incurs a further performance penalty for languages like Dylan.

Perhaps someone should design a DVM from scratch just for Dylan.

-- Brian



Mon, 17 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix


    Date: Thu, 1 Apr 1999 14:34:48 -0800


    >

    >     Date: Tue, 30 Mar 1999 22:34:40 +0200
    >    
    >     Any chance of Harlequin targeting Dylan to a universal platform
    >     like the Java VM?  If so, Dylan could become available on a large
    >     number of CPUs/OSes. I believe the biggest contribution Java made
    >     to software development is not the Java language, but the Java VM;
    >     which unfurtunately is under utilized by programming languages
    >     that compete with Java.
    >    
    > I bet that some of the reasons the Java VM is under-utilized include:
    >  - It's not really very fast.  I hear that even Sun has Java VMs that
    >    run muc faster than the official Java VM.
    >  - It's not very good at supporting higher-order languages (languages
    >    with first-class functions).
    >  - It's got a built-in O-O model that directly supports only single
    >    dispatch in a single-inheritance class model.
    >  - It's got a built-in exception model that supports only exceptions
    >    that "unwind" before any handlers are called.
    > All of these problems can be worked around, but each one of these
    > things incurs a further performance penalty for languages like Dylan.

    Perhaps someone should design a DVM from scratch just for Dylan.

I bet the bought-by-Microsoft Colusa VM would have worked just fine.
Presumably the source code for it is sitting in a box in that huge
warehouse where the Ark of the Covenant is stashed.



Mon, 17 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:
> > -----Original Message-----


> > Sent: Thursday, April 01, 1999 2:35 PM

> > Subject: Re: Harlequin Dylan on Unix

> [...]
> > Perhaps someone should design a DVM from scratch just for Dylan.

> > -- Brian

> IMO, having a Dylan compiler target a VM a waste of an excellent language
> design.

We can agree to disagree then. OCaml is another language with a byecode
compiler, and it is excellent.

Quote:
> It's not that hard to build a dynamic language.

Oh sure, it only takes some of the smartest dynamic language designers
with accumulated decades of experience with CLOS and Smalltalk over
seven years to get just two compilers out.

Quote:
> all to design a VM and target a language to that. What *is* hard is
> combining the benefits of a Dynamic language with efficient compilation to
> native code!

If the language is high level enough, I'm willing to forgive poor
performance, it just means I won't use it for certain tasks. Dylan is IMO
high level enough; I'd be willing to use a moderately fast Dylan with a
good byte code compiler. The existence of such a DVM might make it easier
for implementers to generate dylan compilers, given a widespread DVM.

I wasn't thinking of mobile code, or Obliq, or anything like that.

Quote:
> Java and the JVM are remarkable primarily for what they failed to learn from
> their predecessors.

I don't like Java because it is not only slow (for now) but because it
isn't really that high level a language. It still feels like a neutered
C++. Dylan at least has multi-dispatch and higher order functions.

-- Brian



Mon, 17 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:
> -----Original Message-----


> Sent: Thursday, April 01, 1999 2:35 PM

> Subject: Re: Harlequin Dylan on Unix

[...]

Quote:
> Perhaps someone should design a DVM from scratch just for Dylan.

> -- Brian

IMO, having a Dylan compiler target a VM a waste of an excellent language
design. It's not that hard to build a dynamic language. And it's not hard at
all to design a VM and target a language to that. What *is* hard is
combining the benefits of a Dynamic language with efficient compilation to
native code!

Now, I will admit that mobile code has some cool possibilities. At some
point it would be very interesting to explore the path blazed by the
Modula-3 community there. Modula-3 provides the general-purpose language
with efficient compilation to native code, and an interpreted subset (called
... uhh ... Trestle? Or was that the GUI kit?) is provided to support
applications where code mobility takes precedence over efficiency. And of
course the interpreter (VM) is written in Modula-3, and and can call
arbitrary Modula-3 functions painlessly.

Java and the JVM are remarkable primarily for what they failed to learn from
their predecessors.



Tue, 18 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:
> -----Original Message-----


> Sent: Thursday, April 01, 1999 6:06 PM

> Subject: RE: Harlequin Dylan on Unix

[...]

Quote:
> We can agree to disagree then. OCaml is another language with
> a byecode
> compiler, and it is excellent.

Point taken -- I should have said "having a Dylan compiler target the Java
VM is a waste of an excellent language design." :-)

Quote:
> > It's not that hard to build a dynamic language.

> Oh sure, it only takes some of the smartest dynamic language designers
> with accumulated decades of experience with CLOS and Smalltalk over
> seven years to get just two compilers out.

What I said was:

   "It's not that hard to build a dynamic language. And it's not hard
    at all to design a VM and target a language to that. What *is* hard
    is combining the benefits of a Dynamic language with efficient
    compilation to native code!"

It took all that experience and effort to meet the last goal -- not the
first. If dynamic is *all* you want, there are several *tiny* Scheme
implementations out there. SIOD (Scheme in one defun) certainly qualifies as
a dynamic language in my book -- and I'll bet the original was banged out by
the author in one day.

Quote:
> If the language is high level enough, I'm willing to forgive poor
> performance, it just means I won't use it for certain tasks.

I am sympathetic to this view. In my experience, the marketplace is not.

Quote:
> Dylan is IMO high level enough; I'd be willing to use a moderately fast
> Dylan with a good byte code compiler. The existence of such a DVM might
> make it easier for implementers to generate dylan compilers, given a
> widespread DVM.

> I wasn't thinking of mobile code, or Obliq, or anything like that.

Ah, Obliq! Yes, thank you.

I absolutely agree that a DVM would make it easier to get Dylan running
everywhere -- but at the cost of what makes Dylan worth having everywhere.
Take away the performance edge that Dylan offers over other Dynamic
languages, and you have, well, yet another fine Lisp dialect in a *very*
large crowd of them -- none of which have managed to make serious inroads
into "mainstream" software development to date.

To paraphrase my favorite RFC:

Dynamic, efficient, or easy to implement -- pick any two. Your language
can't be all three.



Tue, 18 Sep 2001 03:00:00 GMT  
 Harlequin Dylan on Unix

Quote:


> [...]
> > We can agree to disagree then. OCaml is another language with
> > a byecode
> > compiler, and it is excellent.

> Point taken -- I should have said "having a Dylan compiler target the Java
> VM is a waste of an excellent language design." :-)

Reading the rest of this reply, we don't disagree. I was suggesting a DVM
as a way to make lots of Dylan environments available. I wasn't suggesting
abandoning native code compilation.

Quote:
> > > It's not that hard to build a dynamic language.

> > Oh sure, it only takes some of the smartest dynamic language designers
> > with accumulated decades of experience with CLOS and Smalltalk over
> > seven years to get just two compilers out.

> What I said was:

>    "It's not that hard to build a dynamic language. And it's not hard
>     at all to design a VM and target a language to that. What *is* hard
>     is combining the benefits of a Dynamic language with efficient
>     compilation to native code!"

It still took a long time just to define the language. So I contend that
language design is hard.

Quote:
> > If the language is high level enough, I'm willing to forgive poor
> > performance, it just means I won't use it for certain tasks.

> I am sympathetic to this view. In my experience, the marketplace is not.

Counterexamples: Perl, Python, Tcl, Smalltalk, ...

All of these have some level of marketplace success. I wouldn't use any
one of them to implement very CPU intensive programs.

Quote:
> > Dylan is IMO high level enough; I'd be willing to use a moderately fast
> > Dylan with a good byte code compiler. The existence of such a DVM might
> > make it easier for implementers to generate dylan compilers, given a
> > widespread DVM.

> > I wasn't thinking of mobile code, or Obliq, or anything like that.

> Ah, Obliq! Yes, thank you.

> I absolutely agree that a DVM would make it easier to get Dylan running
> everywhere -- but at the cost of what makes Dylan worth having everywhere.

No, there is no cost. I didn't say "No native code compilers!" I just said
define a DVM, hopefully a reference implementation, and release it. Mindy
is probably a good place to start.

I like "lightweight" environments to try out new languages. Load OCaml
onto a Windows or Linux box and you can be up and running without a hitch
very quickly on either one. I have done that experiment.

I agree that high performance is essential, and that ultimately a smart
compiler is necessary for that.

-- Brian



Tue, 18 Sep 2001 03:00:00 GMT  
 
 [ 17 post ]  Go to page: [1] [2]

 Relevant Pages 

1. (fwd) harlequin's dylan-corba mailing list switching to MIT's info-dylan

2. Dylan and Java [was: Harlequin Dylan - Update]

3. HARLEQUIN DYLAN and C FFI

4. Small d2c and Harlequin Dylan incompatibility

5. first attempt at Harlequin-Dylan

6. Functional Objects to take over Harlequin Dylan

7. Harlequin Dylan 2.0 beta 2 is now available

8. Future of Harlequin Dylan

9. ODBC database with Harlequin Dylan

10. Harlequin Dylan Update

11. using winInet with harlequin dylan

12. Beginners problem with Harlequin Dylan

 

 
Powered by phpBB® Forum Software