FreeTDS Development Group

[Gregg Jensen <greggj AT savvis.net>]

Kevin wrote:

>         Actually, my boss (who thinks pound defines to be the root of
> all evil) had me rewrite that section without the pound defines, and
> perhaps this is a little cleaner after all...
>
> TDS_SMALLINT tds_get_smallint(TDSSOCKET *tds) {
>   char    buff[2];
>   u_short ms_kludge;
>
>   buff[0] = tds_get_byte(tds);
>   buff[1] = tds_get_byte(tds);
>   ms_kludge = *((short*)buff);
>   return ((TDS_SMALLINT)ms_kludge);
> }
>         However, I think that something is still off in the reads, as I'm
> getting a large number of unknown markers and loosing some information
> on
> the return off the Solaris system still...

Which verision are you working with(VERSION in configure)?  I am using .31 on
Solaris and it seems to be working OK (with your patch to read).  I tried .40 
and
got the results you mentioned.  From one of the earlier messages, someone has 
done
some work on the protocol parsing and it may be off a little bit.  Try the -r 
R0_31
release from CVS (let me know if you need info for this), and see if you have 
better
luck. (you will have to fix the configure, libtool and read.c again though).

Gregg