Mike Thomas | 1 Oct 02:03 2004

RE: Maxima with Corman CL - defsystem defsetf bug

Hi Ray.

| It seems that Corman CL doesn't handle setf quite right. To
| understand get-setf-expansion, get "On Lisp", which has a very nice
| explanation of it (slightly pre-ANSI CL, though).  "On Lisp" is
| available free somewhere on the web.  That certainly helped me a lot
| when I needed to write some setf's sometime ago.

Thanks for taking the time to reply and also for the reference.  It doesn't
seem to mention "get-setf-expansion", but rather "get-setf-method" which
seems to work in a similar way - I'll look further into it.

His view of setf as a function inversion is interesting, and a step beyond
the explanation given in Norvig's excellent "Paradigms of Artificial
Intelligence Programming" which is normally my first CL stuck-grab.

Cheers

Mike Thomas.
James Amundson | 1 Oct 03:03 2004
Picon
Picon

Re: ANNOUNCE: Maxima 5.9.1 released

On Thu, 2004-09-30 at 09:34, Camm Maguire wrote:

> run_testsuite() only shows no errors on the first run.  On all
> subsequent runs, the following three errors appear:
> 
> Error found in /fix/j/camm/maxima-5.9.1/tests/rtest1.mac, problem: (10)
> Error found in /fix/j/camm/maxima-5.9.1/tests/rtest15.mac, problem: (46)
> Error found in /fix/j/camm/maxima-5.9.1/tests/rtest12.mac, problem: (20)
> 
> This has been confirmed at least with latest GCL and clisp.  

Yes. I'm afraid this problem should have been anticipated. It's really a
bug with kill...

> In the second case, I've pinned the non-idempotent statement down to
> the first line of rtest15.mac:
> 
> /* This file contains tests added since April 2002 */
> 
> kill(allbut(props));
> done$

...as you were able to determine. Unfortunately, there isn't an easy
fix. One would like to be able to return maxima to its pristine state
via kill, but it isn't a trivial problem. Although there are bug reports
about kill, this bug should to be in the bug database explicitly.

Thanks for the report.

--Jim
(Continue reading)

James Amundson | 1 Oct 05:00 2004
Picon
Picon

When to open the repository (was Re: First set of cleanups from Andreas Eder checked in)

On Thu, 2004-09-30 at 09:16, Raymond Toy wrote:

> Is the repository open now? 

Your timing is excellent. I was just about to bring up this point on the
list. We only have one major obstacle to opening up the repository to
the developers: downcasing the source. Since downcasing doesn't mix well
with other changes, we have to do it all at once. I can either do it
now, then re-open the repository in a few days, or we can open up the
repository now, then have a very brief close in the next few weeks.

If there are developers with changes ready to commit, we should allow
them to do so before the downcasing. On the other hand, the code will be
easier to deal with once it looks a little more uniform.

What say the developers?

>  I have a couple of small changes to the
> bessel functions to correct an issue where the bessel functions
> weren't being numerically evaluated, and a minor enhancement to allow
> bessel_j to be computed for negative orders.
> 
> There are other numerical evaluation deficiencies there, but I haven't
> fixed those yet.
> 
> I'd also like to unify (maybe) the issue of bessel_j and %j in the
> code.  I think we agreed that all uses of %j should become bessel_j.

It all sounds good.

(Continue reading)

James Amundson | 1 Oct 05:36 2004
Picon
Picon

Re: First set of cleanups from Andreas Eder checked in

On Thu, 2004-09-30 at 09:06, Barton Willis wrote:
> Under windows XP and  gcl 2.6.5 with Andreas Eder's patches,  the test 
> suite finds
> no unexpected errors. 

Great. Thanks.

> Apparently, gcl isn't as picky as sbcl
> 
> $ make clean
> 
> $ make > make.log
> 
> $ grep -ic warn make.log
> 64

Yes. SBCL is really, really picky. That's why I have been using it for
this purpose.

--Jim
CY | 1 Oct 07:03 2004
Picon

Re: When to open the repository

On Thursday 30 September 2004 11:00 pm, James Amundson wrote:
> On Thu, 2004-09-30 at 09:16, Raymond Toy wrote:
> > Is the repository open now?
>
> Your timing is excellent. I was just about to bring up this point
> on the list. We only have one major obstacle to opening up the
> repository to the developers: downcasing the source. Since
> downcasing doesn't mix well with other changes, we have to do it
> all at once. I can either do it now, then re-open the repository
> in a few days, or we can open up the repository now, then have a
> very brief close in the next few weeks.

Will this downcasing be just the files in src, or all the package 
files as well?  Also, will maxima files be downcased now as well?

Personally I'd like to get the downcasing over with, but that's just 
me.

CY
Raymond Toy | 1 Oct 15:21 2004
Picon

Re: Maxima with Corman CL - defsystem defsetf bug

>>>>> "Mike" == Mike Thomas <mike.thomas <at> brisbane.paradigmgeo.com> writes:

    Mike> Hi Ray.
    Mike> | It seems that Corman CL doesn't handle setf quite right. To
    Mike> | understand get-setf-expansion, get "On Lisp", which has a very nice
    Mike> | explanation of it (slightly pre-ANSI CL, though).  "On Lisp" is
    Mike> | available free somewhere on the web.  That certainly helped me a lot
    Mike> | when I needed to write some setf's sometime ago.

    Mike> Thanks for taking the time to reply and also for the reference.  It doesn't
    Mike> seem to mention "get-setf-expansion", but rather "get-setf-method" which
    Mike> seems to work in a similar way - I'll look further into it.

That's why I said it was slightly pre-ANSI.  Also, I think
the get-setf-method he uses doesn't support multiple values, but ANSI
get-setf-expansion does, so there's a few extra things you need to
look at.  At least that's my memory of it.  Using macroexpand on
various setf forms will help you too.

Ray
Raymond Toy | 1 Oct 15:25 2004
Picon

Re: When to open the repository

>>>>> "James" == James Amundson <amundson <at> users.sourceforge.net> writes:

    James> On Thu, 2004-09-30 at 09:16, Raymond Toy wrote:
    >> Is the repository open now? 

    James> Your timing is excellent. I was just about to bring up this point on the
    James> list. We only have one major obstacle to opening up the repository to
    James> the developers: downcasing the source. Since downcasing doesn't mix well
    James> with other changes, we have to do it all at once. I can either do it
    James> now, then re-open the repository in a few days, or we can open up the
    James> repository now, then have a very brief close in the next few weeks.

    James> If there are developers with changes ready to commit, we should allow
    James> them to do so before the downcasing. On the other hand, the code will be
    James> easier to deal with once it looks a little more uniform.

    James> What say the developers?

Up to you.  The changes I have are minor, and are for code that's
already downcased.  I hope, when you do the downcasing and indenting,
that you'll use CL-style if, not emacs if.  (Emacs if-style has the
"then" clause indented one level more than the "else" clause, because
the "else" clause has a implicit progn.)

Ray
James Amundson | 1 Oct 17:01 2004

Re: Re: When to open the repository

On Fri, 2004-10-01 at 00:03, CY wrote:

> Will this downcasing be just the files in src, or all the package 
> files as well?  Also, will maxima files be downcased now as well?

I'm focusing on the files in src. We will do the package files also, but
I think the repository as soon as src is done.

> Personally I'd like to get the downcasing over with, but that's just 
> me.

It isn't just you.

--Jim
James Amundson | 1 Oct 16:59 2004
Picon
Picon

Re: Re: When to open the repository

On Fri, 2004-10-01 at 08:25, Raymond Toy wrote:

> The changes I have are minor, and are for code that's
> already downcased. 

OK. Unless I hear complaints from the other developers in the next
couple of days, I'll do the downcasing, commit, then declare the
repository open.

>  I hope, when you do the downcasing and indenting,
> that you'll use CL-style if, not emacs if.  (Emacs if-style has the
> "then" clause indented one level more than the "else" clause, because
> the "else" clause has a implicit progn.)

Yes. I didn't appreciate this problem when I first started working on
maxima, but I figured it out some time ago. By default, emacs will
indent "if" as follows:

(if (condition)
    (true-case)
  (false-case))

It isn't very common-lisp-like. If you add the following command to your
.emacs:

(setq lisp-indent-function 'common-lisp-indent-function)

the indentation will come out like this:

(if (condition)
(Continue reading)

Raymond Toy | 1 Oct 17:21 2004
Picon

Re: Re: When to open the repository

>>>>> "James" == James Amundson <amundson <at> users.sourceforge.net> writes:

    James> On Fri, 2004-10-01 at 08:25, Raymond Toy wrote:
    >> The changes I have are minor, and are for code that's
    >> already downcased. 

    James> OK. Unless I hear complaints from the other developers in the next
    James> couple of days, I'll do the downcasing, commit, then declare the
    James> repository open.

Would you like help?  There are a ton of files, after all.  I think
the slatec and quadpack stuff is ok, since it's already lowercase.
Indentation might not be quite right, but it's pretty close because I
used cmucl's pprinter to print that code.

Ray

Gmane