Proposal: API for thread-safe time zone functions

Antoine Leca Antoine.Leca at
Wed Jun 13 15:49:28 UTC 2001

Hi folks,

Sorry, I was remote from TZ some days, so I did notice this discussion
until today.
I have not yet read the whole thread, but I should say it looks like
to me a lot of sensible things have been said, so a really good thing
would be to follow someone's, I believe that is Paul's, advice:
a rationale that captures what has been said here would be very welcome,
and this is particularly important for the standarization committees.

This post is to answer an easy one:

Joseph S. Myers wrote:
> * Better time interfaces were listed as one possible item for an amendment
> to C99 (after the problematic changes in some C9X drafts were backed out).
> Antoine Leca may know the current ISO status of this.

Yes. Currently, things are exactly as you describe: this is a possible item,
but not one on which we are currently working on.

If there is some proposal that emerge as being acceptable to the vast majority
of the specialists, then we could move on: that is, designing a "new work item"
for the C standardization committee (and probably or subsequantly the POSIX
committee as well), to design indeed such an amendment, and get it voted in by
the national bodies at ISO that are in charge of programming languages (that is
SC22 if you are OK with ISO hierarchy). Then working as a subgroup including
all of you which are the experts in the field, to issue a formal draft of an
amendment to the C standard. In parallel, we will have to built up implemen-
tations. After that (and we are in the 2004 time frame as a minimum), the baby
will escape us and enter the real process of standardization, with the politics
it involves (this means this can delay things quite a lot).

There have been two major attempts in the (recent) past: one was part of the
C99 process, and it failed because a number of experts did consider that
the proposed changes, while ambitious, failed to solve all the issues.
A lot of the material that is still floating around (and that are conveniently
linked from Markus' page), dated from this epoch.

A second attempt took place as part of the Austin group process (the next
revision of Posix); it was mainly motivated by the 2038 problem (time_t), but
then drifted to more ambitious "solutions", and the subgroup failed to achieve
a minimal level of consensus about the results, so again interest vanished.

While I tried to be as objective as possible, I welcome rectifications.


More information about the tz mailing list