On Wed, 11 Feb 2009, Mindaugas Kavaliauskas wrote:

Hi Mindaugas,

> yes, good idea! BTW, I still do not understood your opinion about internal 
> TIMESTAMP item format: julian and milliseconds, or double julian.fraction?

In the 1-st version I'll use milliseconds as separated member.
We will see how much code we have to update and how it will effect
HVM speed. Later if it's necessary we can change it to double for
better precision. It will not effect existing public C API function
so this modification will be hidden for 3-rd party developers and users.
Just simply the precision becomes bigger. Of course in systems where
time can be taken from OS with such bigger precision, f.e. in Linux.

> And Where can we put timezone offset? How it affects timestamp calculation? 
> I see some hidden problems here.

Now nowhere. I'll introduce it only if we strictly define rules how
it should be used and correspond with OS functions. Without it I see
only possible problems in conversions or comparisons, f.e. storing and
restoring values in database systems which do not support UTC offset
in TIMESTAMP fields. At this moment I do not have enough knowledge to
define all possible conditions and interactions with external code in
all OS-es. For my own use it's not necessary now. I only need some
conversion functions. So I want to leave it to the until we will have
more practice knowledge and feedback from 3-rd party developers.
Of course if some can help here then I can try to already implement it.

best regards,
Przemek
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour

Reply via email to