FreeTDS Development Group

["Peter C. Norton" <spacey-freetds.org AT ssr.com>]

I've isolated the change that breaks pyodbc->unixODBC->FreeTDS->ASE
15.  It seems to be the changes associated with this Changelog entry:

Wed Sep 17 14:14:15 CET 2008    Frediano Ziglio <freddy77_A_gmail_D_com>
        * src/tds/.cvsignore src/tds/Makefile.am src/tds/convert.c:
        * src/tds/data.c src/tds/read.c src/tds/types.pl(added):
        - autogenerate some functions based on types.txt

Can anyone (I'm really hoping that freddy77 has an idea) recommend how
to approach types.txt to fix this?  It looks like the generated
tds_get_conversion_type() is being passed the correct values for the
test columns.  GDB is reporting the last few invocations of
tds_get_conversion_type like this after the change

Breakpoint 1, tds_get_conversion_type (srctype=56, colsize=4) at types.h:124
(gdb) 
Continuing.

Breakpoint 1, tds_get_conversion_type (srctype=56, colsize=4) at types.h:124
(gdb) 
Continuing.

Breakpoint 1, tds_get_conversion_type (srctype=39, colsize=20) at types.h:124
(gdb) 
Continuing.
b
Breakpoint 1, tds_get_conversion_type (srctype=39, colsize=20) at types.h:124
(gdb) 
Continuing.
('01000', '[01000] [FreeTDS][SQL Server]ASE is terminating this process.\n 
(5702) (SQLExecDirectW)')

The first call up the stack is from odbc.c, odbc_populate_ird, and the
second is from odbc_util.c:odbc_set_sql_typeinfo.

I'm hoping that a change to the types.txt will address this problem.
I'm not sure there's enough info in a dump, as I've taken before and
after sessions, and the contents of both are almost completely
identical, except for the memory addresses being passed around.  I
don't know what I'm looking for at this point, so th

I don't understand how to properly manipulate types.txt, or how to
evaluate the correctness of any changes I'm making there, so I'm going
to hold off and keep my fingers crossed that this is specific enough
to get some help, any of which is much appreciated.

Thanks,

-Peter