[Gcl-devel] Re: subtypep

[Camm Maguire]

Greetings!

Henry Baker <address@hidden> writes:

> Camm:
> 
> Which parts of subtypep did you implement?  Did you implement all
> of the stuff re number ranges?
> 

I think we did all of it, at least all the major kingdoms you
outlined.  We are missing 'values come to think of it.  All the number
ranges should be fully implemented, both real and complex. arrays,
cons, singleton, structure, and classes too.  There is some support
for partially unknown type processing, such as 

(subtypep `(and cons (satisfies foo)) 'list)
T
T

The compounding is done via a little generic boolean sigma algebra
code.  You can check out lsp/gcl_predlib.lsp and
ansi-tests/subtypep*.lsp in the cvs head version (2.7.0).  I'm sure
thinks can be further improved, but I'm quite pleased at the
robustness in generic compiler use, especially as measured by the
survival rates in Paul Dietz' random compiler torture tester.

Take care,

> Henry Baker
> address@hidden
> 
> At 02:22 PM 3/7/2006, Camm Maguire wrote:
> >Greetings, and thanks!
> >
> >BTW, also thanks for your paper on a decision procedure for subtypep.
> >It is the basis for GCL's current algorithm, which appears to perform
> >quite competitively despite the implicit exponential behavior on cons
> >types (added to the standard after your paper).  GCL now passes all of
> >the rigorous ansi-tests in this area, which is key its new integration
> >at a deeper level into the compiler.
> >
> >Take care,
> >
> >Henry Baker <address@hidden> writes:
> >
> >> Yes, XML is the "new" S-expression.  It has become the lingua franca of 
> >> "exchange formats" --
> >> e.g., formats of data that have to be exchanged among multiple vendors of 
> >> SW, HW, etc.
> >> For example, "rich text format" ("rtf") used to be the exchange format 
> >> among word processors,
> >> but Microsoft is now pushing a version of XML to take the place of rtf.  
> >> RTF, by the way,
> >> uses braces "{}" instead of parens, and puts a backslash in front of 
> >> symbols, but is
> >> otherwise very lisp-like (I believe that the first version of MS Word -- 
> >> which was done on
> >> the Apple Mac -- was written in Lisp).  Some of the "rtf" files on my web 
> >> site were written
> >> directly in "rtf", rather than being written in Word first, and using Word 
> >> to convert into
> >> rtf.
> >> 
> >> For example, if you have any interest in GPS (Global Positioning System) 
> >> information --
> >> e.g., waypoints, routes, tracks, etc. -- there is a GPX XML format for 
> >> these things that
> >> has become the de facto standard for exchanging this information.
> >> 
> >> At 01:29 PM 3/7/2006, Robert Boyer wrote:
> >> >> Is this useful to other than web programmers?
> >> >
> >> >Beats the hell out of me, but Henry Baker, about whom enough cannot be 
> >> >said,
> >> >seems recently to be of the opinion that xml is the WOTF (wave of the
> >> >future).
> >> >
> >> >True story: A brilliant former UT student of mine, Dan address@hidden, 
> >> >once
> >> >asked me "who is this McCarthy guy and why is he claiming that xml is 
> >> >nothing
> >> >but Lisp?"
> >> >
> >> >Bob
> >
> >-- 
> >Camm Maguire                                            address@hidden
> >==========================================================================
> >"The earth is but one country, and mankind its citizens."  --  Baha'u'llah
> 
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah