proposed changes to zic and zdump
eggert at twinsun.com
Mon Sep 1 04:39:09 UTC 2003
At Thu, 21 Aug 2003 20:37:11 -0400, "Olson, Arthur David (NIH/NCI)" <olsona at dc37a.nci.nih.gov> writes:
> 1. An output file depends on the version of zic that was used to
> create it as well as one or more input files (certainly the file
> that contains the relevant "Zone" line, but also perhaps a leap
> second file and--potentially at least--one or more files that
> contain "Rule" lines. What information is sufficient?
As far as zic goes, the important thing is not the version number of
zic itself; it is the version number of the algorithm that zic uses.
For example, if zic is modified purely to add a new "--version"
option, the algorithm version shouldn't change, since the new option
can't possibly affect the contents of the output files.
> 2. Does the information need to be reflected in the header of the
> output file, or is it sufficient to just append the information to
> the end of the generated output?
The latter, I'd say.
Personally, I'd prefer not to have version information in the output
of zic, as the version info makes it more of a pain to do proper
regression testing. So I'd like zic to have an option to disable this
Better yet, why not add a zic option to enable this new feature, and
leave it disabled by default? That is a more conservative approach,
as it makes it less likely that this new feature will break things.
More information about the tz