Re: compute-system-path needs a call to string to accept keywords for module-name.

[Sam Steingold <sds@gn...>]

> * Marco Antoniotti <znepbkn@...> [2004-02-11 18:46:14 -0500]:
>
> The crux of the problem is FIND-SYMBOL.  You need to change
> FIND-SYMBOL and to attach a case-sensitive flag to each package. It
> ain't perfect, but it's much better than any other solution.

<http://clisp.cons.org/impnotes/pack-dict.html>;:
MAKE-PACKAGE accepts a keyword argument :CASE-SENSITIVE. Similarly,
DEFPACKAGE accepts an option :CASE-SENSITIVE. When its value is non-NIL,
the package will be case sensitive, i.e., the reader will not
case-convert symbol names before looking them up or creating them in
this package. The package names are still subject to (READTABLE-CASE
*READTABLE*), though.

-- 
Sam Steingold (http://www.podval.org/~sds) running w2k
<http://www.camera.org>; <http://www.iris.org.il>; <http://www.memri.org/>;
<http://www.mideasttruth.com/>; <http://www.honestreporting.com>;
Lisp: its not just for geniuses anymore.

compute-system-path needs a call to string to accept keywords for module-name.

[Pascal J.Bourguignon <pjb@in...>]

While compiling lisa, I get this error:
*** - MAKE-PATHNAME: illegal :DIRECTORY argument (:RELATIVE :LISA)

which occurs in defsystem-3.x/defsystem.lisp (cvs 1.66):

(defun compute-system-path (module-name definition-pname)
  (let* ((file-pathname
	  (make-pathname :name (etypecase module-name
				 (symbol (string-downcase
					  (string module-name)))
				 (string module-name))
			 :type *system-extension*))
         (lib-file-pathname
	  (make-pathname :directory (list :relative module-name) ;;<--- here
                         :name (etypecase module-name
				 (symbol (string-downcase
					  (string module-name)))
				 (string module-name))
			 :type *system-extension*))
         )
 
valid pathname directory  n. a string, a list  of strings, nil, :wild,
:unspecific, or some other object  defined by the implementation to be
a valid directory component.


It  seems that  module-names  are expected  to  possibly be  keywords,
therefore I'd suggest to use string here:

<	  (make-pathname :directory (list :relative module-name) ;;<--- here
>	  (make-pathname :directory (list :relative (string module-name)) 


-- 
__Pascal_Bourguignon__                     http://www.informatimago.com/
There is no worse tyranny than to force a man to pay for what he doesn't
want merely because you think it would be good for him.--Robert Heinlein
http://www.theadvocates.org/

Re: compute-system-path needs a call to string to accept keywords for module-name.

[Marco Antoniotti <marcoxa@cs...>]

Thanks for the report.  It was fixed in 1.67.

However, on the side, I have noted that there is an abuse of keywords 
in the CL practice for things like this (and package naming).  My 
diagnosis goes back to THE BIG MISTAKE (with the power of hindsight :) 
) made when CL was made uppercase default, and to the kludgy and non 
backward compatible hack by Franz.  Using strings is not a bad idea, 
especially when you need to name things that eventually become part of 
a pathname.

Cheers

marco



On Tuesday, Feb 10, 2004, at 04:11 America/New_York, Pascal 
J.Bourguignon wrote:

>
> While compiling lisa, I get this error:
> *** - MAKE-PATHNAME: illegal :DIRECTORY argument (:RELATIVE :LISA)
>
> which occurs in defsystem-3.x/defsystem.lisp (cvs 1.66):
>
> (defun compute-system-path (module-name definition-pname)
>   (let* ((file-pathname
> 	  (make-pathname :name (etypecase module-name
> 				 (symbol (string-downcase
> 					  (string module-name)))
> 				 (string module-name))
> 			 :type *system-extension*))
>          (lib-file-pathname
> 	  (make-pathname :directory (list :relative module-name) ;;<--- here
>                          :name (etypecase module-name
> 				 (symbol (string-downcase
> 					  (string module-name)))
> 				 (string module-name))
> 			 :type *system-extension*))
>          )
>
> valid pathname directory  n. a string, a list  of strings, nil, :wild,
> :unspecific, or some other object  defined by the implementation to be
> a valid directory component.
>
>
> It  seems that  module-names  are expected  to  possibly be  keywords,
> therefore I'd suggest to use string here:
>
> <	  (make-pathname :directory (list :relative module-name) ;;<--- here
>> 	  (make-pathname :directory (list :relative (string module-name))
>
>
> -- 
> __Pascal_Bourguignon__                     
> http://www.informatimago.com/
> There is no worse tyranny than to force a man to pay for what he 
> doesn't
> want merely because you think it would be good for him.--Robert 
> Heinlein
> http://www.theadvocates.org/
>
>
> -------------------------------------------------------
> The SF.Net email is sponsored by EclipseCon 2004
> Premiere Conference on Open Tools Development and Integration
> See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
> http://www.eclipsecon.org/osdn
> _______________________________________________
> clocc-list mailing list
> clocc-list@...
> https://lists.sourceforge.net/lists/listinfo/clocc-list
>
--
Marco Antoniotti					http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group		tel. +1 - 212 - 998 3488
715 Broadway 10th FL				fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.

Re: compute-system-path needs a call to string to accept keywords for module-name.

[Drew McDermott <drew.mcdermott@ya...>]

>  ... THE BIG MISTAKE (with the power of hindsight :) 
> ) made when CL was made uppercase default, and to the kludgy and non 
> backward compatible hack by Franz.  

Non-backward-compatible, yes.  Kludgy, no.  They've just undone the
big mistake.  It is not hard to write one's code so it runs in both
Ascii mode and Franz's "modern" mode.  I wish CLOCC code had this
feature, but I recognize that my wish is unlikely to be granted.

The problem with modern mode is that if you actually use it you have a
big problem importing someone else's code.  Nonetheless, the "big
mistake" is so big and annoying that I'm willing to go through some
amount of hassle to undo it.

-- 
                                   -- Drew McDermott
                                      Yale Computer Science Department

Re: compute-system-path needs a call to string to accept keywords for module-name.

[Marco Antoniotti <marcoxa@cs...>]

On Wednesday, Feb 11, 2004, at 18:34 America/New_York, Drew McDermott 
wrote:

>
>>  ... THE BIG MISTAKE (with the power of hindsight :)
>> ) made when CL was made uppercase default, and to the kludgy and non
>> backward compatible hack by Franz.
>
> Non-backward-compatible, yes.  Kludgy, no.  They've just undone the
> big mistake.  It is not hard to write one's code so it runs in both
> Ascii mode and Franz's "modern" mode.  I wish CLOCC code had this
> feature, but I recognize that my wish is unlikely to be granted.
>
> The problem with modern mode is that if you actually use it you have a
> big problem importing someone else's code.  Nonetheless, the "big
> mistake" is so big and annoying that I'm willing to go through some
> amount of hassle to undo it.

Well, yes.  But there are other ways to fix the problem in a far less 
intrusive and much more portable way.
The crux of the problem is FIND-SYMBOL.  You need to change FIND-SYMBOL 
and to attach a case-sensitive flag to each package. It ain't perfect, 
but it's much better than any other solution.

Franz did not do that and failed to propose something acceptable.  
Maybe not kludgy, but unacceptable yes.

Having said that I have often wondered what was in the mind of the 
people doing the ANSI when they decided on making CL uppercasing.  It 
will never stop from baffling me.

Cheers

marco


--
Marco Antoniotti					http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group		tel. +1 - 212 - 998 3488
715 Broadway 10th FL				fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.

Re: compute-system-path needs a call to string to accept keywords for module-name.

[Sam Steingold <sds@gn...>]

> * Marco Antoniotti <znepbkn@...> [2004-02-11 18:46:14 -0500]:
>
> The crux of the problem is FIND-SYMBOL.  You need to change
> FIND-SYMBOL and to attach a case-sensitive flag to each package. It
> ain't perfect, but it's much better than any other solution.

<http://clisp.cons.org/impnotes/pack-dict.html>;:
MAKE-PACKAGE accepts a keyword argument :CASE-SENSITIVE. Similarly,
DEFPACKAGE accepts an option :CASE-SENSITIVE. When its value is non-NIL,
the package will be case sensitive, i.e., the reader will not
case-convert symbol names before looking them up or creating them in
this package. The package names are still subject to (READTABLE-CASE
*READTABLE*), though.

-- 
Sam Steingold (http://www.podval.org/~sds) running w2k
<http://www.camera.org>; <http://www.iris.org.il>; <http://www.memri.org/>;
<http://www.mideasttruth.com/>; <http://www.honestreporting.com>;
Lisp: its not just for geniuses anymore.

Re: compute-system-path needs a call to string to accept keywords for module-name.

[Marco Antoniotti <marcoxa@cs...>]

Yes.  I know and I did take into account when writing up the thing.  
CLisp is the closest to my proposal.

Marco


On Thursday, Feb 12, 2004, at 08:47 America/New_York, Sam Steingold 
wrote:

>> * Marco Antoniotti <znepbkn@...> [2004-02-11 18:46:14 -0500]:
>>
>> The crux of the problem is FIND-SYMBOL.  You need to change
>> FIND-SYMBOL and to attach a case-sensitive flag to each package. It
>> ain't perfect, but it's much better than any other solution.
>
> <http://clisp.cons.org/impnotes/pack-dict.html>;:
> MAKE-PACKAGE accepts a keyword argument :CASE-SENSITIVE. Similarly,
> DEFPACKAGE accepts an option :CASE-SENSITIVE. When its value is 
> non-NIL,
> the package will be case sensitive, i.e., the reader will not
> case-convert symbol names before looking them up or creating them in
> this package. The package names are still subject to (READTABLE-CASE
> *READTABLE*), though.
>
> -- 
> Sam Steingold (http://www.podval.org/~sds) running w2k
> <http://www.camera.org>; <http://www.iris.org.il>; 
> <http://www.memri.org/>;
> <http://www.mideasttruth.com/>; <http://www.honestreporting.com>;
> Lisp: its not just for geniuses anymore.
>
--
Marco Antoniotti					http://bioinformatics.nyu.edu
NYU Courant Bioinformatics Group		tel. +1 - 212 - 998 3488
715 Broadway 10th FL				fax. +1 - 212 - 998 3484
New York, NY, 10003, U.S.A.