FreeTDS Development Group

[Frediano Ziglio <freddyz77 AT tin.it>]

Il gio, 2002-08-29 alle 18:36, Brian Bruns ha scritto:
> tds_process_row() is getting unwieldy.  Post 0.60 work should include
> streamlining this func.  Looks like we can simply test for
> column_unicodedata and instead call tds_get_string instead of tds_get_n. 
> I don't have access to SQL Server where i am now to test this.  Did you
> break the switch out to a separate function?
> 
> Brian
> 
I fixed problem with ntext (and some other tsql related).
I not added any function and not simplified tds_process_row.
One reason is to not add possible unseen bugs.
Another is this related to ucs2 coding. I think in next release to
improve two think:
- internalization
- odbc driver
We use for i18n an early conversion method (conversion from ucs2 to byte
charsets just after reading from network). I think best method is a late
conversion (convert just before returning data to client). So we should
mantain ucs2 coding in tds and converting (for example) in cs_convert.
This should solve problem with utf8. Also doing so is possible to change
coding in the middle of the dialog.
This is why I propose to remove unicodedata flag...

freddy77