FreeTDS Development Group

[Alex Kiesel <alex.kiesel AT document-root.de>]

On Mon, 2004-01-26 at 11:24, ZIGLIO, Frediano, VF-IT wrote:
> > 
> > On Mon, 2004-01-26 at 10:08, Alex Kiesel wrote:
> > > Next to my patch, I'm sending a unittest for ctlib (t0010.c), which
> > > simulates what I did manually with sqsh.
> > 
> > t0010.c seems to have been stripped out of my mail, you can find it
> > here:
> > http://document-root.de/tds/t0010.c
> > 
> 
> Good. I note you set timer before ct_command/ct_send. Perhaps it's
> better to set time after sending command. Why you do not use alarm
> instead of setitimer (alarm should be more portable...). I'll add test
> to CVS (renamed and with some TODO note).

You're probably right; I used setitimer, because `man alarm` says:
DESCRIPTION
     This interface is made obsolete by setitimer(2).
[...]

> I look at your patch. On ATTN you just set a flag however server do not
> known you want to cancel current command. It would be better if you send
> cancel to server too.

Ok, you could immediately call tds_send_cancel(), set the flag and call
tds_process_cancel() when the main applications tries to read the next
time.
This would match the ct-reference description even more exactly.

Greets,
-Alex