[tz] Recent conversions of zones to aliases

David Patte ₯ dpatte at relativedata.com
Fri Oct 3 01:23:38 UTC 2014


On 2014-10-02 19:34, Stephen Colebourne wrote:
> Just to note my continued unhappiness with the conversions of zones to aliases.
> The recent change to map Harare to Maputo broke a test in Joda-Time
> (the most widely used 3rd party library in Java for date/time,
> probably used by millions of developers). This means that the many
> users of Joda-Time will get errors when they try to recompile the
> latest jar file, a process that has worked fine for years. In order to
> make the test pass I now have to special case to say that either
> Harare or Maputo is a valid value.
> This issue occurs because Joda-Time normalises all aliases to the
> actual zone as per TZDB data. Thus a request for
> DateTimeZone zone = DateTimeZone.forID("Africa/Harare");
> which use to return a zone with the ID "Africa/Harare" now returns
> "Africa/Maputo". This means that users can no longer get a zone that
> represents their country for half of Africa and the Carribean.
> I cannot stress enough that these utterly unecessary changes have and
> will continue to cause observable differences in end-user
> applications. I remain firmly of the opinion that there should be at
> least one actual Zone entry for each ISO defined territory/country,
> and that these recent changes should not be described as minor
> tidying.
> Stephen


More information about the tz mailing list