Re: POP3 Audit

[Peter Bloomfield <PeterBloomfield MindSpring com>]

On 2001.10.05 20:46 Ali Akcaagac wrote:
> oki last one...
> 
> i was able to track things down to till end was a bit of work
> but exactly here it fails. ive tried several things and also
> placed some gdb outputs to the draft box but then i deleted
> them because i was able to track the problem down until reaching
> this point...
> 
> gtk_signal_emit(GTK_OBJECT(mailbox), libbalsa_mailbox_signals[CHECK]);
> 
> then bummer... but something i dont understand is, check_mailbox_list
> will also be called without pthreads support so it also goes into
> libbalsa_mailbox_check ... this tells us that the problem must
> have happened before...

[gdb transcript snipped]

Hi Ali!

I admire your tenacity in tracking down this problem!

I occasionally see what I believe to be another instance of the same 
general problem, but associated with IMAP transactions. Balsa starts, 
but exits a fraction of a second after displaying the main window. If 
gdb is running it, the status is, of course, `program exited normally'.

The problem is *much* more frequent when I have a slow connection to 
the IMAP server (> 500 ms ping time, for instance), so I infer that 
it's related to long blocked reads. I've never seen it on checking my 
POP3 server, but the ping time there is always around 20 ms--do you 
ever have delays with yours?

I also have the *impression*, not backed up by serious observation, 
that if Balsa fails in this way, and I restart it immediately, it 
starts successfully, which suggests to me that something outside Balsa 
is involved--some condition that gets cleared by the fail. However, I 
may be fooling myself there...

One last clue: a while ago, before some upgrade (and there've been so 
many!), I used to get a console message `program received real-time 
signal 32' when this type of failure happened; now there's nothing.

I hope some of these recollections help in tracking down the 
problem--best of luck!

Peter