FreeTDS Development Group

["James K. Lowden" <jklowden AT schemamania.org>]

I've put the finishing touches on the new row buffering for db-lib.  Like
any good rewrite, it took three tries....  It's pretty major surgery,
because dblib's row buffering logic is always active; the degenerative
case is the normal one, with only one row buffered.  

Consequently, I'm looking at the unit tests pretty closely.  They all pass
but #8, and for the life of me I don't see how t0008 *ever* worked.  But
just in case I'm missing something obvious, I thought I'd ask someone else
why a program that ends unconditionally with:

        fprintf(stderr, "dblib failed for %s\n", __FILE__);
        free_bread_crumb();
        return 1;

wouldn't always look like a failed test?  Here's the end of the output:

Read a row of data -> 45 row 045   
Read a row of data -> 46 row 046   
Read a row of data -> 47 row 047   
Read a row of data -> 48 row 048   
dblib failed for ../src/dblib/unittests/t0008.c
FAIL: t0008

I don't see how it can be avoided.  I think the test is broken.  

Otherwise, I plan to commit the new buffers tomorrow.  

--jkl