[cvsnt] Re: timezone to use with update -D option?

[Gerhard Fiedler]

Tony Hoyle wrote:

>> It seems that update -D requires local time rather than UTC. I suspect it's
>> the client's local time, but haven't had a chance yet to test this. 
> 
> I'm not sure.. the date parser is inherited from the old cvs (and 
> predates even that I think).

Wow... I only remember having read in everything I read about cvs that
dates/times are in UTC. All of them... I don't remember having read
anything about local time being used /anywhere/. But according to you this
seems to have been in cvs/cvsnt for a loooong time. Did I just miss this in
the docs, or is this really strange?


>> Maybe this is the result of having a system developed on the Greenwich
>> meridian... it might have been safer to put the UTC meridian somewhere in
>> the Pacific :)

> Not my fault!! :)

Probably not, I was just kidding :) 

But did this never come up during testing? I mean, I don't use the update
to date option a lot (in fact, almost never). But the moment someone uses
it (which probably happens during testing) who's not on UTC, it is pretty
obvious that there's a difference between the date/time sent with the
command and the resulting sticky date. Yet I don't remember having read
about this in any documentation of cvs/cvsnt. 

You probably have a test case for the update -D option in your test suite,
don't you? What does this test case say about this?

Gerhard