FreeTDS Development Group

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

Hi All.

I've made some changes and am checking them into CVS currently.  Among the
changes is better handling of data type conversions. src/dblib/convert.c
is changed (hopefully for the better), adn I think all the dbconvert crap
will go easier now.  

Also I changed the mechanism for passing column values from tds to the
upper layers.  This was done mostly to support row buffering, specifically
with an eye towards getting PHP to work, since it uses row buffering.
Anyway, the old mechanism was just to have a 255 char field in the
TDSCOLINFO struct and the upper layers looked for it there.  This was
especially wasteful on memory since even a 1 byte value would get 255
characters.  (Actually this was a temporary fix that stayed far too long).
The new mechanism is to calcuate the row size when the result set is
processed and pack all the values into one buffer. This will work
especially well for row bufferring as we will need only one memory
allocation for the entire buffer instead of mallocing for every row or
column.

Sorry for being long winded, but the point of all this is that dblib
seems to be working ok, but Ive definately broke ctlib for the time being.
I will try to get it back in working order soon.

If any of the PHP guys on the list are reading, if you have a way of
testing the modules (ie sybase) without a webserver (maybe some test 
harness code) please let me know it would help shorten the debug cycle.

Brian