svn commit: r274365 - head/contrib/tzcode/stdtime

[John Baldwin]

On Tuesday, November 11, 2014 12:47:40 pm Andrey Chernov wrote:
> On 11.11.2014 19:34, John Baldwin wrote:
> > On Tuesday, November 11, 2014 11:06:24 am Andrey Chernov wrote:
> >> On 11.11.2014 18:34, Garrett Cooper wrote:
> >>>> This code is wrong in any case, setting EOVERFLOW even in unneeded
> >>>> no-error cases like breaking from the loop. See how NetBSD does it
> >>>> correctly instead.
> >>>
> >>> Hi!
> >>>     I saw your reply earlier but hadn't gotten an opportunity to revise 
> > the patch like you suggested. I'll look at the NetBSD code and see if I can 
> > adapt it to our copy of tzcode.
> >>> Thank you!
> >>>
> >>
> >> Currently as we need only EOVERFLOW parts from it, I don't think it will
> >> be hard task to merge only them.
> > 
> > Do you know if it is fixed upstream?  I want to import newer tzcode from 
> > upstream to get implementations of the extensions to allow use of arbitrary 
> > time zones (mktime_z, etc.)
> > 
> 
> If you mean https://github.com/eggert/tz some NetBSD fixes, including
> part of EOVERFLOW fixes are merged there by Christos Zoulas reports, see
> the history in
> https://github.com/eggert/tz/commits/master/localtime.c
> But not all of them, NetBSD have more places with EOVERFLOW set (still
> missing in upstream). F.e. time2sub EOVERFLOW checks are not merged yet.
> BTW, latest NetBSD version claims that they merge 2014i tz code, so have
> mktime_z, etc too.

Yes, I mean that as the upstream.  Sounds like merging in updated tzcode would
be compatible with both of our goals (though more EOVERFLOW fixes might be
needed afterward).

-- 
John Baldwin