On 12/4/17 21:04, Tom Lane wrote:
Michael Douglass <mikeadouglass@gmail.com> writes:
I'd suggest a first step would be to provide a unique - reasonably short - opaque id for each definition. That allows us to build out a localization framework around it. This strikes me as a remarkably bad idea. Reality is that the existing tzdb zone names are a de facto part of the universe now. Whether you think that they ought to be used in UIs or not, *they are so used*, and you aren't going to get anywhere by telling people not to.
I can positively guarantee that if those names are replaced by meaningless IDs, the very first thing that the Postgres project will do is build a mapping table between those IDs and the old names, and continue to expose the old names in our UI. I confidently predict that a majority of other consumers of the tz database will do likewise. There's nothing to stop people using those names if they so wish - it's just a simple mapping. I suggested that such a mapping table should be provided. The point is to move the arguments over names out of this project. Also to stop misleading people into believing the current tz names and aliases have any official meaning.
There are a number of organizations which are much more appropriate for handling naming.
regards, tom lane