Re: signature terminology (editorial BLD issue)

Naming discussions can quickly go askance.  I don't have any strong feelings on 
this at all, especially since its sort of a hidden-from-users (important for 
implementors) thing in RIF.

Still, I think the reason you find the naming unnatural is that you are 
accustomed to non-polymorphic signatures. In that simple case, "signature 
expression" and "signature" are the same thing (every signature has one 
expression).  It is in the case of polymorphic signatures that there is a 
difference, and I do prefer the name signature the way it is used in the spec, 
since a signature should be the acceptable (allowed) type mappings for a symbol. 
  It is the *symbol* that has a signature.

-Chris

Sandro Hawke wrote:
> Chris Welty <cawelty@gmail.com> writes:
>> Jeez, I'm sure glad I didn't have Michael for a professor.
> 
> I'll avoid making any comments about my time having Chris as a teacher
> (he was a grad student, not a prof), since I dropped the class after a
> couple of weeks.  :-)
> 
>> The differences are pretty simple, though:
>>
>> A signature has a name and a set of expressions.
>> A signature name is just a symbol used to reference the set of expressions.
>> A signature expression is the standard sort of thing you might think of as a 
>> signature
> 
> Right -- that's the part that bugs me.  ("Okay, here we have apples and
> oranges and bananas.  Let's call apples, "bananas".  Now, hand me a
> banana.")
> 
> Here are names that match my intuition:
> 
>      (i) => bool                     a signature 
>      { (i i) => i, (i) => bool }     a signature set
>      MySig                           a signature set name
>      MySig{(i i) => i, (i) => bool}  a signature block ?
>                                      (a signture set with its name)
> 
> A "signature block" assigns the name to the set and has a "return value"
> of being the set, right? Is it worthwhile combine them?  How about just
> having assignment, and using the name, later....
> 
>     MySig = {(i i) => i, (i) => bool}    a name assignment
>                                         
> 
> My problem with "expression" is that it's usually a general term for any
> linguistic construct.  Those things above are all linguistic expressions
> in the signature language, so they all seem like "signature
> expressions".   Maybe some of you can keep track of when a banana is a
> banana and when it's an apple, but that's more work than I want to do
> unless it's really needed.  
> 
>      -- Sandro
> 

-- 
Dr. Christopher A. Welty                    IBM Watson Research Center
+1.914.784.7055                             19 Skyline Dr.
cawelty@gmail.com                           Hawthorne, NY 10532
http://www.research.ibm.com/people/w/welty

Received on Tuesday, 25 September 2007 20:47:36 UTC