Обсуждение: PQfinish kills my app

Hi all, I have a small app that is dying whne PQfinsih() is called.  

My app has two main options, -v and -a.  It dies on -a, not -v.

The app will copy users mailfile and store this in the database.  So my
options are -v for verbose output, and -a for all users.  When run as
normal or -v it checks the users id, and calls function get_user_mail,
then exits via an exit_nicely function (straight from the sample progs).

When run with -a (for all) it runs function get_users_mail() which is a
simple loop that calls get_user_mail() for each user, then exits with
the same exit_nicely() function.

So I don't understand why it dies on PQfinish one way but not the
other.  But, if I do not call exit_nicely (ie: just end without
PQfinish) ir dies on option -v but works for option -a

PS: By die I mean seagfault, I traced wit with gdb to the PQfinish
function.  As I am new to the GNU tools (and fairly new to C) I am not
sure whats going on.  I could post the code, but was not sure about the
list rules, it's 8 k long.

Thanks


-- 
 Rob Brown-Bayliss ---====<*>====---
 live from ZOOstation http://www.ZOOstation.cc

Rob Brown-Bayliss <rob@ZOOstation.cc> writes:
> So I don't understand why it dies on PQfinish one way but not the
> other.  But, if I do not call exit_nicely (ie: just end without
> PQfinish) ir dies on option -v but works for option -a

Kinda sounds like you might be calling PQfinish more than once on
the same connection object.  That's a no-no.

> PS: By die I mean seagfault, I traced wit with gdb to the PQfinish
> function.  As I am new to the GNU tools (and fairly new to C) I am not
> sure whats going on.  I could post the code, but was not sure about the
> list rules, it's 8 k long.

People have posted much larger messages than 8k with less reason.
If the multiple-calls idea doesn't pan out, let's see it...
        regards, tom lane