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

On Thursday, Feb 12, 2004, at 11:31 America/New_York, John Foderaro 
wrote:

>
>>> You can have it both ways, (a) or (b)
>>>
>>> 	(symbol-name 'cl:car) ==> "CAR"
>>>
>>> or
>>>
>>> 	(symbol-name 'cl:car) ==> "car"
>>>
>>> What you want is
>>>
>>> 	(find-symbol "cAr") ==> #S(SYMBOL :NAME <"CAR" or "car" or "Car" or
>>> ... > :package "CL")
>>>
>
>
> You keep bring up find-symbol.  At this point I don't care about 
> find-symbol.
> All I want to know is what symbol-name returns.  It's a string
> and I might want to use that string to create other strings.
> Are you saying that my program can't know what symbol-name returns
> since that depends on the package name case and if I'm writing
> some utility function I can't know ahead of time what the 
> package-name-case
> of each package will be when this function is loaded in to lisp?

In a certain sense yes.  For CL you, as an implementor, have the 
liberty of choosing if you want lowercase of uppercase or mixed case 
symbols.  As a user I do not want to care.  I will just know that the 
CL package is a "handle with care" package.  If you want you can also 
add an extra property for packages with PACKAGE-SYMBOL-NAME-CASE set to 
:IGNORE that records what is the original setup.  In the case of the CL 
package you can have :UPCASE or :LOWERCASE (or something else)

>>> (note: no dynamic
>>> *package-symbol-name-case* available)
>
> are you saying that you have to decide the package-symbol-name-case
> at system build time?

You need to decide at package definition time.  When you call 
DEFPACKAGE or MAKE-PACKAGE.

> In that case you're going to have to either have it set to make
> ANSI code that you might want to import happy or you're going
> to have to have two images (which is the thing about the ACL solution
> that you don't like).  Either way you can't get by with exactly
> one Modern image.

No. For legacy packages I assume that the PACKAGE-SYMBOL-NAME-CASE will 
be set to :IGNORE.  That is why FIND-SYMBOL is crucial here.  For *new* 
packages you have full control.

What you are asking is whether you can do

	(setf (package-symbol-name-case some-package) <some agreed upon key>)

I think this should be restricted by establishing rules for "upgrading".

As I said, my proposal is not fool proof.


>
> I may have misunderstood things because again I haven't seen
> the whole proposal.

The proposal is in the thread I sent you.  Doing a Google on 
SYMBOL-NAME-CASE will bring it up.  Eventually I will go through the 
exercise of writing it up.  I would be happy if anybody suggested an 
avenue for publication :)

>
>>> ... and I still am wondering what were you guys thinking while 
>>> writing
>>> the standard (you were there, weren't you?) :)
>
> This is a sensitive subject to me.  I argued strongly for a
> Modern Lisp (though it wasn't called Modern then).
> At the time we had Franz Lisp which was a Modern Lisp yet it
> was able to make use of large libraries of ANSI-type code (including
> Macsyma) without changing a line of the imported code.
> So there was no issue of leaving Lisp code behind if we looked
> ahead and saw that Modern mode was the wave of the future.
>
> However the case-insensitive crowd won and I can't say why
> without disparaging a few people who are still around and I
> don't want to do that.

... allllriiight! :)   Let's not go there then.  :)

I am sorry, but right now I really do not have the time to go on on 
this.  I actually should have not brought it up in the first place.  At 
this point it is just a purely academic exercise for me.  I will try to 
clean up the proposal and put it up somewhere.  But not today or next 
week :)

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.



-------------------------------------------------------
SF.Net is sponsored by: Speed Start Your Linux Apps Now.
Build and deploy apps & Web services for Linux with
a free DVD software kit from IBM. Click Now!
http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click