On Dec 16, 2014, at 4:17 AM, Lester Caine <lester@lsces.co.uk> wrote:
tzdist is a working group of IANA currently trying to formalise an API to allow TZ data to be supplied as a service. There are demo servers but nothing has been formally published yet.
There is a bit of a sticking point at the moment as to whether the service needs to worry about the 'version' of TZ data served up, and the original draft simply assumed that the latest version of TZ is all that is needed, but I have a problem with that since material published today using today's version of TZ may be broken if a later version is used without any means of identifying the fact.
“without any means”? But there is a means, that’s why all well-designed protocols have version numbers. Any valid tzdist proposal would have to include the tzdata version designation as part of what it provides. There is a separate possibility that some tzdist client might get confused by a new tzdata entry, but that’s mostly the client’s issue, and possibly tzdata, it has nothing to do with the distribution protocol. Or are you arguing that tzdist should be a mechanism for distributing obsolete data, alongside current data? Clearly it could be, but I would be surprised if there were significant interest in that. paul