FreeTDS Development Group

[Frediano Ziglio <freddy77 AT gmail.com>]

2014-02-27 20:35 GMT+00:00  <Burkhard AT habmalnefrage.de>:
> (resending as plain text - used HTML before which seemed to cause problems)
>
> Hello,
>
> A few years ago (time flies), I got a very helpful suggestion from this 
> list on how to patch freetds 0.82 in order to handle UTF-16 characters in 
> BMP-1 correctly (i.e. characters outside the UCS-2 two byte range). See 
> http://lists.ibiblio.org/pipermail/freetds/2010q3/026240.html .
>
> This has been working well for me since then. Now, I need to upgrade to 
> 0.91, and I'm seeing that the problem still exists. There have been some 
> changes to iconv.c in the meantime. Is it still safe to make the changes to 
> 0.91 in a way similar to what you suggested back then for 0.82?
>
> In this quarter's mailing list archive, I'm seeing that just a few days ago 
> you have been discussing changes to the iconv related code in freetds. 
> Maybe this could be a good time also to integrate a change related to this 
> problem?
>
> Best Regards,
>   Burkhard
>
>
>
> Gesendet: Donnerstag, 27. Februar 2014 um 21:23 Uhr
> Von: Burkhard AT habmalnefrage.de
> An: freetds AT lists.ibiblio.org
> Betreff: [freetds] SQL Server UTF-16 characters with FreeTDS 0.91

If system iconv support every type of conversions there is no problems
replacing ucs-2 with utf-16 (beside it's hard to predict output length
in all cases). The problem is that some iconv implementations support
conversions only to/from ucs-2 or utf-8 (surely hp-ux one). The idea
is to force using an implementation (like GNU iconv) that does not
have these issues.

Frediano