[Pan-users] Re: Re: Unable to post to news groups

[Duncan]

Andy posted <address@hidden>, excerpted below,  on
Sun, 07 Sep 2003 19:45:20 +0100:

> On Sunday 07 September 2003 16:31, Christophe Lambin wrote:
> 
>> Not as long as it should be: I only see you retrieving new headers here.
>> The actual post appears to be missing.
>>
>> Could you try again?
> 
> Hi Duncan
> Thanks for getting back to me. I ran the debug script to append to a file and 
> then posted and - wouldn't you guess !! - I could post again!!! I'm not sure 
> what the heck this is - but it is feeling a bit like when one takes a noisy 
> car to a mechinc and it doesn't make the noise at the required time!!!
>
> Since it is working I won't submit the file, but should that problem
> start again, then I definitely will come back and get some more input.
> Thanks for your input thus far Duncan.

Well, I'd love to claim credit, but I only responded once to this thread
(until now, anyway <g>), and that wasn't of much help except to say I'd
had a problem with posting b4 so could sympathize, and remind you that the
PAN version number would be helpful.  After that, Charles posted the
"debug sockets" suggestion, and Chris posted the quoted above.

Just in case you don't recognize the names, check PAN's about box, author
tab.  <g>  Me?  I know enough "developer-speak" to understand most of
what's discussed and can read code well enough to follow the idea, at
least (when well commented, anyway, as the PAN code I've looked at is),
but I don't know C well enough to be an effective programmer.  However, I
help out on the list where I can, and report bugs where I can, and
occasionally make my opinion known, where I think I need to..  and..
occasionally make hopefully educated comments like the following, when I
feel inspired to do so.

There's one class of bugs that's notoriously hard to track or debug, thus,
hard to squash.  Unfortunately, your description of a bug that disappears
with attempts to debug, fits the description to a "T".  These bugs have to
do with threading and thread resource or mutex (mutually exclusive
section, only one thread can enter it at a time) contention, because it
often happens that under debugging, the process is slowed down enough by
the debugging that the conflict works itself out -- the threads align
differently because the timing is different, and there's not a problem,
only to have the bug reappear when the program is run normally again.

Now, far be it from me to know whether this fits the mold, as I'm writing
this without even as much as looking at the code to see if threading is
even involved in this section, tho it must be to some degree as the post
would be made on one of the connections out of several possible, but
that's often the first thing that pops into a programmer's head when they
have a multi-threaded app, and they read about a problem that "magically"
disappears in debug mode! <g>

.. Just so you know what might be going on if the bug continues to
trouble you in regular mode, but it just will NOT occur in "debug sockets"
mode. .. To bad the noise that never occurs for the mechanic doesn't have
a similar absolutely logical explanation!  <g>

-- 
Duncan - List replies preferred.   No HTML msgs.
"They that can give up essential liberty to obtain a little
temporary safety, deserve neither liberty nor safety." --
Benjamin Franklin