CTO down again.

[Brian Rice]

On May 15, 2004, at 1:41 PM, Francois-Rene Rideau wrote:

> I think it's a problem with araneida being too prone
> to jumping into the debugger instead of just ignoring failures
> in HTTP transaction.

That should be pretty much the fix needed. I'm not sure any more how or 
where to go about doing this for Araneida, so if it's not immediately 
obvious, just bounce this back to Dan Barlow and see what he thinks. 
His usual recommendation is to run it behind apache to filter out the 
nonsense, which we can do, but really shouldn't be necessary. (I like 
reminding him that it /does/ run quite well without a proxy, and maybe 
just such a patch would make it perfectly obvious why.)

> (And yes, I sometimes forget to leave my attachtty sessions.
> I suppose you can kill the attachtty, or maybe attachtty again
> and detachtty will close connection to the previous attachtty.)

Actually, attachtty just does that for you, which may be a little hairy 
if someone's actively working in it already, but looking at the process 
table lets one know quickly enough whether that's the case (and I 
needed to do that to see what socket to grab).

Incidentally, all I did was to attach and poke around, eventually just 
hitting ^C to break out of the debugger. Pretty simple, really, which 
is why a handler should be simple and worthwhile.

--
Brian T. Rice
LOGOS Research and Development
http://tunes.org/~water/