On 5/16/12 8:13 AM, "Jonny Lamb" <jonny.l...@collabora.co.uk> wrote:
> On Fri, 2012-05-11 at 09:42 -0600, Peter Saint-Andre wrote: >> We do have a spec for "Entity Time"... >> >> http://xmpp.org/extensions/xep-0202.html > > Sorry, I had actually seen this, I should have referenced it in my first > mail. The problems I see with the XEP-0202 approach and the advantages > of adding a 'timezone' element to the XEP-0080 are: > > * The XEP-0202 approach requires you to poll all your contacts as there > is no change notification. That's a bug, which is easy to correct. Imagine that XEP-0202 says you can use PEP with the node name urn:xmpp:time. > * Having both location and timezone information in one PEP node > guarantees atomicity. For example, it's easy to get into a situation > where the XEP-0202 timezone reports Europe/London and the location PEP > node reports you're in San Francisco. There are two questions I like to ask myself when I'm wondering if data belongs in a single PEP node: 1) is the information semantically coherent? Does it *mean* the same thing? 2) does it change at the same rate? Timezone changes less frequently than location. In this case, it's not much data, so that's not of paramount importance. Timezone is usually thought of as more-or-less dependant on location, but I can imagine cases where I set my personal timezone to somewhere other than where I'm sitting. E.g. Read "Eastern Standard Tribe" by Cory Doctorow ( http://craphound.com/est/). > * If you are publishing your location anyway in the XEP-0080 PEP node > then it's one less thing to change, from the perspective of the client > implementers and network traffic. But it also allows clients that don't care about your timezone to opt out. Either way, it's not *that* much data, particularly if you've got XEP-0312. > One advantage of XEP-0202 over the XEP-0080 approach is that the latter > will not work on Google's servers. If we keep adding value to PEP, either Google will have to implement it, or people will want to switch away from their service. > How about: said new 'timezone' element in the XEP-0080 PEP node which > cross-references XEP-0202 in an attempt to keep them in sync? That's better than adding new elements in the location namespace in -80, I guess. -- Joe Hildebrand
