Use TZMetaData to identify what type of field

[miab3]

Hi Jan,

Of course you can test on Linux, maybe right away in the 21c Express Edition.

In fact, I hardly use Oracle, and my attention to the problem was that I came across it during testing.
I dealt with it temporarily by using the table names explicitly.

One more thing.
Note that the server can be on Linux and the client can be on Linux but also on Windows

Michał

[aehimself]

I changed the driver to return the CHAR_LENGTH value for string types in the COLUMN_SIZE column and to only use DATA_LENGTH for the CHAR_OCTETS_LENGTH column.

You da man, Jan. Yep, works like a charm, finally could get rid of my dummy query :)

Opinions?

Yes, don't ever roll back this commit, please! :)

With all due seriousness, though...

The current calculation only makes sense for IDEs that use ANSISTRING and not UNICODESTRING as their default string type...

This information is coming from the database, not the IDE. If I need the buffer size and I know all drivers return it as ANSI... I'll have no issues having a magic number multiplier hardcoded.

[marsupilami]

The current calculation only makes sense for IDEs that use ANSISTRING and not UNICODESTRING as their default string type...

This information is coming from the database, not the IDE. If I need the buffer size and I know all drivers return it as ANSI... I'll have no issues having a magic number multiplier hardcoded.

That is the problem - they don't give consistent results currently