On 11/6/05, Tim Bray <[EMAIL PROTECTED]> wrote: > > <co-chair-mode> > There are a flurry of Paces, and there is enough camera-ready spec > text supporting all the options that we believe that they are well- > understood by the group. > > For a good introduction, see Rob Sayre's note at http://www.imc.org/ > atom-protocol/mail-archive/msg02834.html > > The co-chairs feel that WG consensus in favor of hierarchical > drilldown or date-ranges is unlikely. That leaves the other two > options, which, it seems, are sort of independent. That leaves the > WG with two choices to make: > > 1. Should we use URI templates? (the current draft only has {index}, > and it seems unlikely that we could get consensus on anything more > ambitious).
-0.5. I can see the potential power of a URI templating mechanism but just to support paging it seems like overkill. In other words, this has been scaled down to the point of where it isn't justified; this being said I also wouldn't justify scaling it up for core use cases. Perhaps better done as an extension that could ultimately support more than just paging. > 2. Should we have prev/next pointers in collections to allow clients > to page through? +1, I think giving the server the option to do chunking for large collections, and having a well-defined client model for accessing the data when chunked is important. > On the face of it, it seems like all four possible outcomes are > plausible. > - No templates, no paging -1, need some solution for large collections > - Templates but no paging -0.5, see above > - No templates but paging +1, see above > - Both templates and paging -1, hard pressed to see why you need two solutions to the same problem.
