Re: around the table on datatypes [ was: Re: datatyping draft 3 (for telecon)]

Sans Chapeau

I normally try to remain neutral on technical discussions, but I have a 
serious concern about the datatyping proposal which I feel duty bound to 
draw to the attention of the WG.
So I'll reply to my own message (I seem to have been talking to myself more 
than usual lately)

In doing so, I'm aware that I suggested that I believed that a synthesis of 
the two proposals was possible, and various folks have worked very hard and 
done a great job to create that synthesis.  However, having seen the 
result, I think we've got a camel, not the arabian thoroughbred I hoped 
for.  I was wrong.

At 19:34 18/02/2002 +0000, Brian McBride wrote:
>At 23:58 14/02/2002 -0500, Pat Hayes wrote:
>>Latest version of the datatype summary document now available at
>>
>>http://www.coginst.uwf.edu/users/phayes/DatatypeSummary3.html
>>
>>incorporating ideas arising from discussions with Patrick S. (rdfs:drange 
>>and especially section 10).
>
>I would like to "go round the table" of the WG on the latest datatype 
>proposal.  By go round the table, I mean to solicit the views of each 
>member of the WG, without initiating a debate on members views.  We can 
>then summarize those views and deal with issues arising.
>
>Please answer the following questions:
>
>   o Does the datatyping proposal meet your
>     needs and the needs of your users?
>     (Who are they?) (What is missing?)

- No.  I am hearing that users require S-B and that seems very plausible to me.

My users are users of the Jena toolkit.  I'm not hearing from them 
directly, but basing this view on what I'm hearing from folks in more 
direct contact with users.



>   o Are there features that could be dropped and
>     still meet the needs of your users? (Which?)

Yes.  I believe both the datatype triple (S-A) and doublet (TDL/S-P) should 
be dropped.
S-B is sufficient for now.  User experience and would tell us if S-B alone 
was inadequate, at which time, and with greater understanding of user 
needs, further machinery could be added.  As we have discovered, it is 
harder for a WG to withdraw a feature from a spec than it is to add to 
it.  Better to only put in what you know is needed, and add further stuff 
later as necessary.

I strongly believe we must drop at least one of the doublet or datatype 
triple mechanisms.


>   o Does the proposal 'work for you'?

No.


>   o Are there any concerns with the proposal
>     you would like to raise? (What are they?)

My main concern is complexity.  This proposal is too complex for our users 
and this will significantly affect its uptake and use.

Patrick's quiz was a very useful contribution and a went a long way to 
convincing me of this.  I've been tracking the datatyping discussion for a 
while.  I've enjoyed long talks about the issues with Jeremy.  I read the 
latest DT document and then tried Patrick's quiz.  I failed.

It was too hard, which tells me that this datatyping proposal is too hard 
and too complex for our users.

Looking at it I see two ways of doing effectively the same thing.  The 
datatype triple approach is simpler, so I prefer that.  Adding the doublet 
approach adds as a lot of complexity and I don't see that, in practise, I 
get anywhere near enough value from that extra complexity.  I have found 
the suggestions so far for the benefits of adding the doublet approach 
unconvincing.

RDF/XML syntax has received much criticism for providing more than one way 
to say something.  But at least, once the syntax has been parsed into a 
graph, there is only one graph (give or take some issues with BagID).  The 
current datatype proposal, with both the triple and the doublet approaches, 
allows two different graph structures to represent the same 
information.  This is a bad idea.  Not everyone is programming in prolog or 
the like where a simple rule can transform one into the other.

In summary, I believe that S-B on its own is sufficient for now.  If the WG 
feels strongly that a richer mechanism is needed then it should choose just 
one.

Having expressed my concerns, I do not feel, that as co-chair, I can 
champion them in the WG.  If others are sympathetic to the views I have 
expressed, they must take on that task.

And now with my co-chair hat back on, I'd remind folks we are out of 
time.  We have to choose between the options we have; we don't have time to 
go back again over old ground.

Brian

Received on Wednesday, 20 February 2002 06:59:48 UTC