Formats of tz-data & docs
Oscar van Vlijmen
ovv at hetnet.nl
Thu Apr 20 12:56:45 UTC 2000
Just a couple of thoughts in the discussion regarding the formats of tz-data
Maintainers and users of the excellent tz-code and -data: please note that
you do not live in a UNIX-only world. There are some people who like to
produce and already have produced timezone products, based on Windows
applications, Mac OS applications and web interfaces.
It seems not very useful and even difficult to compile or port the tz-code
to other operating systems.
The tz-data however is very useful to everybody.
In my opinion the current format of the tz-data is OK and should not be
changed. Every change in the format would mean that the parser had to be
rewritten and rigorously tested. Too much trouble with hardly any gains at
the moment; for UNIX users that is. Or am I missing something?
The tz-data could however be formatted a bit more consistently. Sometimes
the fields in Rule and Zone lines are separated by spaces, in most cases by
tabs and sometimes by nothing more than a seemingly arbitrary field width.
All field data should be consistently separated by tabs. This would enhance
the 'importability' to other platforms.
The documentation regarding the tz-data should be in a platform independent
format. Non-UNIX users can do virtually nothing with the formats n/t/groff,
Texinfo, DocBook, linuxdoc-sgml, whatever.
One of the most useful formats at the moment would be HTML 4.0.1.
In short I suggest:
* Don't XML-ize or MIME-ize or VTIMEZONE-ize at the moment.
* 'Make' or convert to HTML all documentation; to be added to the current
* Clean up the Zone & Rule formatted data.
Oscar van Vlijmen
In case of replies: please use my usual address o.van.vlijmen at tip.nl
More information about the tz