FreeTDS Development Group

[Brian Bruns <camber AT umcc.ais.org>]

On Mon, 8 Feb 1999, Gregg Jensen wrote:

> OK,
> Maybe, this will be the last time about this particular item.  The problem
> turns out to be on the caluclation of more_results when the marker is
> TDS_END3_TOKEN.  The output of the get_incoming function gave
> me the thought that maybe the byte that it was looking for was reversed
> (endianness stuff).  Well it looks like that may be the case.  At least,it
> fixed the two queries I was having problems with (and the other queries
> run too).  The sp_help tablename dumped core in sqsh, but I think it was
> due to a different problem (float conversion, which still may be an
> endianness problem).
> 
> 
> This is what I did in tds_process_end:
> 
>  unsigned char  tmp = tds_get_byte(tds);
> 
> #if defined(HW_BIG_ENDIAN)
>    tmp = tds_get_byte(tds);
>    *more_results = (tmp & 0x1) != 0;
>  tds_unget_byte(tds);
> #else
>    *more_results = (tmp & 0x1) != 0;
> #endif
> 
> instead of just:
> 
> *more_results = (tmp & 0x1) != 0;
>
this sounds like it is a 16bit field, maybe better to call
tds_get_smallint() on it and let that deal with it. Then just shorten the
subsequent call to tds_get_n from 7 to 6 bytes.

> There are two places where this exists, the one mentioned above and
> in tds_process_default_tokens (look for '& 0x1').  This is a first cut
> so please take pot shots at this to make it more production worthy.
> 
> Gregg
> 

Brian