axiom-developer
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Axiom-developer] CCL maintenance.


From: Stephen Wilson
Subject: Re: [Axiom-developer] CCL maintenance.
Date: Wed, 30 May 2007 18:30:06 -0400

From: Stephen Wilson <address@hidden>
Date: 30 May 2007 18:30:29 -0400
In-Reply-To: <address@hidden>
Message-ID: <address@hidden>
Lines: 123
User-Agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.4
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
--text follows this line--
"Bill Page" <address@hidden> writes:

> On May 30, 2007 12:48 PM Stephen Wilson wrote:
> > ...
> > Bill Page wrote: 
> > > As part of the open source release NAG (Mike Dewar) claimed that
> > > they had delivered Axiom to Tim Daly in a form that would compile
> > > using CCL on Linux. I have never tested this and I think it would
> > > be good to know since it at least gives us a standard benchmark
> > > for comparison.
> > 
> > I assume you mean comparison against other Lisp hosts? Im curious
> > about the benefit in that.
> 
> I meant "benchmark" against which to compare the mathematical output
> of Axiom. I think it is important to be able to test to make sure
> that the changes we make to the open source version do not lead to
> any feature reversions compared to the commercial version of Axiom.
> As I understand it there still are a few people and places that
> have this version. Otherwise I don't think there is much value in
> just comparing a CCL-based version of Axiom to implementations in
> other Lisps.

Ok, thanks for explaining.  As Tim pointed out though there are
currently better ways of doing this.

> > It would be one thing if Axiom was targeting one Lisp implementation
> > in an attempt to exploit all of the extensions and non-standard
> > features that particular Lisp provides.  But in this case it would
> > most assuredly not be CCL.
> > 
> > On the other hand, targeting multiple Lisps is of main benefit to
> > the user (and a non-negligible burdon on the Axiom developers).
> 
> I don't see what benefit targeting multiple Lisps has for end users
> of Axiom. To me this seems of benefit only to the developers.

The basic issues in my mind is portability, performance, and to be
capable of distributing axiom without a lisp attached (as we do with
GCL).  If i were new to axiom I would love to go to the website,
download the source, compile it with the sbcl I have installed, and
have a working system.  

But of course, there are some benifits for developers.  Notably being
able to choose the host lisp with which to work.

> > ... 
> > Unfortunately I cant really comment on this aspect.  I have not
> > used windows for many, many years.
> 
> I am sorry that you are so isolated from the mainstream... ;-)

Im leaving for a fishing trip to the back country in a few hours, so
Im going be isolated even further :)

> > I assume what your saying is that CCL provided a means to emulate
> > the missing unix functionality on windows without the need for
> > a compatibility layer like msys or cigwin?  I could see how that
> > would be an advantage for the average windows user as they are
> > only interested in a binary blob that works for them.
> 
> Yes.
> 
> > We would need to figure out how to effectively cross-compile axiom
> > into a native windows app.  I dont see any of that happening myself
> > so I would simply rely on my clouded perspective that all the world
> > is unix, and let msys/cigwin take care of the rest.
> 
> ??? There is no version of Axiom (or GCL) for cygwin yet. But
> Axiom already compiles as a native Windows app user MSYS/MinGW.

Ok, forgive my ignorance on this.  I thought those environments gave
access to something approximating a full POSIX system (which defines
utilities) even during runtime.

> > > Other things that might be for some use in CCL relate to the
> > > "saturn" browser interface for Windows which unfortunately was
> > > not part of the open source release but might be important for
> > > future attempts to rivive this kind of interface on Windows
> > > and Linux.
> > 
> > I never saw "saturn", so i dont really know what it is. I have
> > no way to resurect it or test it.  I doubt I will spend time
> > reverse-engineering something like that.
> 
> Too bad. I think it was a thing of great beauty that was lost
> to the world due to the proprietary software development model.
> 
> > 
> > Besides, I thought the whole Axiom-in-a-browser was the 
> > preferred route towards a cross-platform interface?
> 
> Yes. That is precisely what I am talking about. That is what
> "Saturn" was. Please refer to the Axiom developer email
> archives.

Ok. Ill have a look.  There has been a lot of discussion about the
browser UI on the list the last few years so I must have missed the
saturn connection.

> > ... 
> > Initially, supporting CCL would not be terribly expensive.
> > I belive the build amchinery is present in all the branches to
> > support it.  Its the long term cost of having to maintain both
> > Axiom _and_ CCL that is convincing me the resources are not
> > available. Esp. since Axiom wants to be Ansi compliant and CCL
> > is certanly not.
> >
> 
> It is a major assumption to say that "Axiom wants to be Ansi
> compliant". Obviously this depends on how you view the relationship
> between Axiom and Lisp.

I view axiom as a Lisp program.  As soon as you start tracking a deep
bug in Axiom, this fact is overwhelmingly obvious.

ANSI defines the basic shape shape of the language for me, as for most
Lisp developers and compiler writers.  Its only natural that we move
to converge on this standard.


Sincerely,
Steve





reply via email to

[Prev in Thread] Current Thread [Next in Thread]