Re: FLD Review part 0.1

Michael Kifer wrote:
>> I suggest this be slightly weakened to use "should" instead of the implied 
>> "must", ie "...dialects should be derived from RIF-FLD by specialization..."
> 
> "Should" means "you are welcome to ignore," which is not the intent here.
> Of course, "must" also means nothing, since there is no provision for RIF
> police to enforce it. Nevertheless, I strongly believe that it should be a
> "must" to emphasize the intent.

I strongly disagree, and suspect others would feel more strongly about it.

"Should" does NOT mean "you are welcome to ignore," even if you interpret it 
that way.  It means, formally, that all complying implementers must in good 
faith try to obey the requirement or document why they cannot.  In this case, we 
are expecting the implementers to be in W3C working groups, and thus they have a 
very strong criterion to provide this documentation.

In standards-speak, it is the word "may" that is interpreted as you take it.

Again, because of the way the phrase is currently worded, it comes down to 
whether or not future dialect designers want to call their dialect a "logic" 
dialect (FLD says it constrains all "logic dialects"). So, currently, if a 
dialect designer wants to ignore FLD, they just remove "logic" from the name of 
the dialect. Thus I think by wording it too strongly you are actually 
discouraging people from trying to use FLD to the extent possible.

So, in fact, I would like to strengthen the potential applicability of FLD.  I 
am (now) suggesting we word it so that all dialects *should* specialize FLD, to 
the extent they feel is possible, not just because they desire the moniker 
"logic" in their dialect name, but because they wish to be RIF dialects and 
encourage interoperability.  We saw this in the PRD effort, and while I'm sure 
it was painful, I think trying as hard as they could to be closer to BLD (and 
thus FLD) led to a greater degree of interoperability.

-Chris

>> I can imagine some group wishing to say they have a "logic" dialect even though it is 
>> incompatible with FLD.
> 
> Incompatible (rather than requiring an extension)? I cannot imagine that.
> A logic rule language like that? Why should we be bothered with such an
> improbable eventuality?
> 
> There are languages, like Mercury or Prolog, that are logic-looking, but
> incompatible and not fully based on a logic. If somebody wants to call
> such a language "logical" then, as far as RIF is concerned, that category name
> is already taken. So they would have to invent a different category.
> 
> This is a much cleaner solution, and it also helps prevent confusion.
> 
> michael
> 
> 
> On Thu, 14 May 2009 23:07:19 -0400
> Chris Welty <cawelty@gmail.com> wrote:
> 
>> Working on my FLD review.  In general I don't think any of my comments are that 
>> major or will require any discussion, except potentially:
>>
>> In the abstract and overview, I think the standard-speak used is too 
>> constraining regarding the requirement of future dialects to implement FLD:
>>
>>  >Abstract: "The actual dialects are required to specialize this framework to 
>> produce their syntaxes and semantics. "
>>
>>  >Overview: "All logic RIF dialects are required to be derived from RIF-FLD by 
>> specialization, "
>>
>>
>> I suggest this be slightly weakened to use "should" instead of the implied 
>> "must", ie "...dialects should be derived from RIF-FLD by specialization..."
>>
>> In general, the text is pretty good about saying that FLD may need to be updated 
>>   in light of new dialects, but I still think our message is that FLD *should* 
>> be used, not *must* be used, in the specification of new dialects.
>>
>> Of course one could argue that FLD is only making this requirement of dialects 
>> that wish to be called "logic" dialects, but as with the naming of "not", which 
>> came down to sort of "who gets the right to call their negation 'not'", I can 
>> imagine some group wishing to say they have a "logic" dialect even though it is 
>> incompatible with FLD.
>>
>> -Chris
>>
>>
> 
> 

-- 
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, 26 May 2009 14:45:30 UTC