Hi Alexey,--On November 22, 2006 2:30:59 PM +0000 Alexey Melnikov <alexey.melnikov@xxxxxxxxx> wrote:
My comment to Dan on this was that it might make sense to have a way to "profile" ANNOTATE so that servers could offer limited portions of it. e.g. a server might choose not to implement either SORT'ing or SEARCH'ing on annotations and that could be indicated via ANNOTATE-NOSORT and ANNOTATE-NOSEARCH capabilities.I can see a reason for ANNOTATE-NOSORT, but why ANNOTATE-NOSEARCH? This is just making something already complex even more complex.
Its up to implementers to decide whether supporting search on annotations is hard or not and whether they want the option to not do it. I'm not saying we should allow that, but if there are reasons to then we can easily do so.
e.g. if a server wanted to support a limited set of entries, then a new 'ALLOWEDENTRIES' option to SELECT could be used to return the set of supported entries on a mailbox (e.g. just "/comment" in Dan's example) - this could also cover support for .priv and .shared. The benefit of doing this, rather than inventing a new extension with new syntax, is that it offers an upgrade path to full annotate. It also means that anyone implementing full annotate on a client will be able to easily handle "less capable" servers by simply looking at what is supported. I think this is better than having two extension. This could be done as a follow-up to the current "experiment".+1
-- Cyrus Daboo