- From: Erik Wilde <dret@berkeley.edu>
- Date: Wed, 25 Jun 2008 10:46:03 -0700
- To: Chris Butler <cbutler@dash.net>
- CC: public-geolocation@w3.org
i think the problem with this kind of "semantics-altering optional parameter" is that i would bet that in 99.9% of implementations, when i use a date and specified "CALSCALE:AZTEC", they would still treat all dates as gregorian. Chris Butler wrote: > Hi everyone. > > One option would be to do something similar to what the iCal RFC did > with calendar types: > > http://tools.ietf.org/html/rfc2445#section-4.7.1 > > In the iCal case only Gregorian was ever used, but I believe the thought > was that eventually you could have Julian, Hijiri, Hebrew, Aztec, etc. > > In the geo case there may be only one for now (WGS84) but it allows for > extension of that... > > Thanks. > > Chris > > -----Original Message----- > From: public-geolocation-request@w3.org > [mailto:public-geolocation-request@w3.org] On Behalf Of Erik Wilde > Sent: Wednesday, June 25, 2008 10:00 AM > To: public-geolocation@w3.org > Subject: Re: The Geolocation API must provide location data in terms of > a pair of latitude and longitude coordinates. > > > Jon Ferraiolo wrote: >> For lay people (such as me) who until now have never heard the term >> "geodetic system" before, perhaps the API can have the ability to > return >> a simple lat/long value but somewhere inside the spec there can be an >> explanation of which geodetic system (or whatever the proper term is) > is >> used by default. > > it seems that wgs84 more or less is the de-facto standard for > lat/long values these days, but that definitely needs to be specified. > if you do have a gps, it very likely has some configuration screen where > > you can choose from a number of reference systems. this is important > when you want to look up coordinates on a paper map: if the gps > reference system and you map's coordinate system do not match, it is > very hard to actually locate the coordinates on the map (the > transformations can be fairly complex). > > as for elevation values, the reference system is much more important. > devices can and do use different reference systems, or they may even use > > barometric elevation measurements (which tend to be more precise, > depending on the number of satellite fixes), so for the elevation to > make any sense, there definitely needs to be a reference system. and > because it is entirely reasonable for a device to use gps values for > lat/long and barometric values for the elevation, the reference system > cannot be just one for the complete location, it must be specific for > these values. > > and like i mentioned earlier, and somehow going in the same direction, i > > am very interested in the API supporting URI-based locations. in a way, > this is just another "reference system", where instead of lat/long > values, the location is defined by a URI scheme. like i said, i am not > very happy with the current geo: scheme proposal (in particular the > latest version of it, which switched to a tile-based scheme), but > narrowing down location to lat/long only is a rather limited location > concept. > > in particular, location URIs can provide a good solution for come of the > > privacy issues of the location API. URI-based location may use > coordinate-based lat/long locations (such as proposed by the current > geo: scheme draft), but they could also represent locations providing > better privacy, such as city names, states, or entirely user specific > locations, which still make sense in the context of interactions among > closed user groups (e.g. on social networking sites), but have far fewer > > privacy implications than gps-precision coordinates. > > cheers, > > dret. > -- erik wilde tel:+1-510-6432253 - fax:+1-510-6425814 dret@berkeley.edu - http://dret.net/netdret UC Berkeley - School of Information (ISchool)
Received on Wednesday, 25 June 2008 17:47:04 UTC