14.06.2019 16:17, Adriano dos Santos Fernandes wrote:
On 13/06/2019 17:05, Vlad Khorsun wrote:
13.06.2019 16:44, Alex Peshkoff via Firebird-devel wrote:
On 13.06.2019 12:43, Vlad Khorsun wrote:

First, you loose things. The adjusted (displayable) timestamp is not
convertible back for duplicated timestamps (DST end).

   Not sure i got you. Could you provide an example ?


At 03:00:00 due to DST change time is set to 02:00:00. To what UTC
corresponds 02:30:00?

   It is fully depends on region name and date part of timestamp with
TZ, but...

It seems you not read my answer to the end.

   User pass to the server some timestamp and specify in what TZ it was
set.
At least user think he pass data in this form.

   Why we return back UTC instead of original region timestamp ? Why
conversion
to the original (region) value must be done by client ? What benefits
for client
it have ?

Do you even worked with a library doing datetime logic in a way suitable
for multiple time zones?

Every one stores the UTC time more the region _or_ offset. The
displayable value is calculated only when one wants to display the value.

  This is exactly what i want. The only difference with current state
of things is to move that calculation from client to the server - at
the moment when it returns\receives data to\from client.

Every one who tries to store adjusted time fails, including:

  I not offer to store adjusted time.

1) Me, in first tries to create the feature
2) Windows, which by default does it (in relation to hardware clock) and
cannot even dual boot correctly with others OS

To store adjusted datetime for a region-based value, it's also needed to
store the offset (in other words, to store the adjusted datetime it's
also necessary to store the UTC datetime), otherwise one cannot convert
it back to UTC. We surely does not want a wrong data structure for
datetime values informed by user before flow from client to server.

I will stop repeating myself explaining the same thing again and again.

  Better re-read what i wrote.

Regards,
Vlad



Firebird-Devel mailing list, web interface at 
https://lists.sourceforge.net/lists/listinfo/firebird-devel

Reply via email to