TIME / TIMESTAMP with timezone and fractional seconds about nanodbc

[lexicalunit]

I've very much for this! When I first created the time and timestamp structs I didn't have a good use-case for them because I personally was not actually using them in my code. So I have never felt entirely confident that the structures as-is are sufficient for the needs of others. I'm quite sure that timezone, at the very least, will be a very useful addition.

[lexicalunit]

From @mloskot on November 16, 2016 17:33

I'm glad.

My main use case is around SQL Server, but hopefully those changes will allow to cover PostgreSQL too, and others if necessary.

By the way, despite SQL Server Native Client (SNAC, SQLNCLI.H) is not supported beyond SQL Server 2012, the new Microsoft ODBC Driver for SQL Server still reports TIME and DATETIMEOFFSET SQL types using constants as defined in SQLNCLI.h as explained in Data Type Support for ODBC Date and Time Improvements

Time    SQL_SS_TIME2    -154 (SQLNCLI.h)
DatetimeOFFSET  SQL_SS_TIMESTAMPOFFSET  -155 (SQLNCLI.h)

It's a bit of a mess, but we may need to accommodate some changes for SQL_SS_TIME2 and SQL_SS_TIMESTAMPOFFSET too. I'm going to check what constants report pgsqlODBC and others which support similar types.

[lexicalunit]

From @mloskot on November 21, 2016 10:18

My last https://github.com/lexicalunit/nanodbc/issues/224#issuecomment-261014201 turned into a separate issue #226

[lexicalunit]

From @mloskot on November 21, 2016 15:15

PR #228 adds very minimal support to handle the SQL Server-specific types with existing nanodbc::timestamp, without adding any new structures yet. The latter part is still being discussed.

[lexicalunit]

From @mloskot on November 21, 2016 17:5

Following my findings in #229, extended nanodbc::time would not be useful in case of PostgreSQL data types mapping.

I suspect, given the current knowledge, the extended nanodbc::time|timestamp structures would be only dedicated to SQL Server-specific uses. If that is true, perhaps it is better to introduce separate structures and document their uses (or even conditionally-compiled for better clarify, as discussed in https://github.com/lexicalunit/nanodbc/issues/226#issuecomment-261984633).

[lexicalunit]

From @mloskot on November 22, 2016 21:59

I started thread on pgsqlODBC list: Binding time and timestamp with time zone data type

I also tested and confirmed, what was actually expected, that binding timestamp column as SQL_C_SS_TIMESTAMPOFFSET instead of SQL_C_TIMESTAMP does not work with the pgsqlODBC driver.

[mloskot]

@mkaes I think this one can be closed too by #272 ?