[anaphora-devel] aand only does the first argument?

[Gary King]

Hi Nikodemus,

I agree with you but I wasn't aiming for flexibility. I was proposing  
to use the *features* as a way to migrate from one semantics to  
another b/c IMHO, aand should be like the one in On Lisp that binds  
`it` all the way down and aand1 should be the one that binds `it` to  
only the first clause. However, it sucks to break downstream code  
willy-nilly.

I suppose what one really wants is "magic" that lets you remap symbols  
when you use a package (similar in spirit to what Eiffel lets you do  
when you do subclassiing... (n.b., my Eiffel knowledge is _way_ out of  
date) or maybe something that mimics "darcs replace" to changes tokens  
(only with proper language support so that things _never_ break (I did  
say magic, remember!).

In this case, I'm happy enough with aand* and this is a much simpler  
solution. In the general case, I'm not sure if there are good general  
solutions but it's fun to think about them.

thanks,

On Jun 8, 2008, at 9:32 AM, Nikodemus Siivola wrote:

> On Sun, Jun 8, 2008 at 3:58 PM, Gary King <gwking at metabang.com> wrote:
>
>>> Please no. Using compile time *FEATURES* to change semantics like  
>>> that
>>> is pretty horrible.
>>
>> Well then, a defvar?
>
> Let me rephrase that: having any sort of magic that changes the
> semantics of an interface is pretty horrible.
>
> Granted, sometimes the line between "changing semantics" and "having a
> flexible interface" is a fine one.
>
> Maybe I'm missing something here, but what you propose seems to be
> pretty squarely on the "changing semantics with a magic toggle" side
> of the line. If we just add AAND*, then I don't see the need for the
> toggle.
>
> Cheers,
>
> -- Nikodemus

--
Gary Warren King, metabang.com
Cell: (413) 559 8738
Fax: (206) 338-4052
gwkkwg on Skype * garethsan on AIM