- From: Graham Klyne <gk@ninebynine.org>
- Date: Tue, 30 Sep 2003 10:43:49 +0100
- To: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Cc: w3c-rdfcore-wg@w3.org
At 11:36 29/09/03 +0100, Jeremy Carroll wrote: >I agree that these are possible weaknesses, I think the schedule issue is >an important one. Please suggest improvements - possibly to do with use of >this document, and a separate rebuttal that consists of >a) the blow-by-blow account I added in today's draft >b) discussion of wider issues such as schedule and no compelling reason >for delay Hmm... I should have expected this (which is a quite proper and reasonable response). Rather than attempt a new drafting, what I'm going to do here is suggest an outline drawing upon existing material that I hope might present the information in a way that more closely reflects (my perception of) this group's reasons for not acceding. [[ Why we didn't accept I18N's objection to the design of XML literals 1. Review status of the WG, noting that we're long overdue, losing participation, and that the I18N desideratum for "seamless plain and XML literals" was only articulated well after the end of LC1. This might use material from my earlier message [1]. 2. point by point response to I18 concerns, per Jeremy's message [2], in particular the new final section of [3]. 2a. note that the WG has wrestled long and hard with datatype issues (which are closely related, and in particular that the handling of both XML and language tags is intimately bound up with the MT for literals). Due to conflicting desiderata, there are no easy fixes that don't require a lot of reworking. [Cite multiple early designs for datatyping] 2b. Reference to an appendix with Jeremy's discussion of some alternative designs, noting that this isn't exhaustive but illustrative of the technical difficulties we face and also that we did consider alternatives. This is by way of response to the I18N claim that they offered alternative solutions. 3. Note that many of the difficulties we have faced in clarifying the earlier M&S design were caused in part by ill-considered late additions to that specification, and the consequent implementation practices that have grown up as a result [cite examples?] 4. Summary: in the absence of a compelling claim that anything is actually broken (i.e. despite prompting for actual cases where the current solution is inadequate, none has been offered) we feel that it is more important to deliver what we have than to risk failure to deliver anything at all. A point to draw out here is not to claim that alternative technical designs are not possible, or maybe even better than what we have, but to note that the effort of producing one may be too much for this WG and it's allotted time. Appendix: Jeremy's discussion of some possible solutions, from [2] References to supporting material ]] My hope would be that the main response would be no more than a page, or a couple of screens of text, and the detailed technical arguments are available through references to supporting material. Does this help? #g -- [1] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Sep/0276.html [2] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Sep/0275.html [3] http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2003Sep/att-0275/i18n-part.html ------------ Graham Klyne GK@NineByNine.org
Received on Tuesday, 30 September 2003 06:05:02 UTC