On Wed, 19 Nov 2014, Nicholas Boel wrote to mark lewis:

 ml> actually, it is a problem on both ends of that intermediate 
 ml> connection...

 ml> firstly, the sending system should strip off certain attributes 
 ml> when the message is packed... attributes like CRAsh and IMMediate, 
 ml> especially... DIRect possibly but if DIRect was set, the message 
 ml> should not be on any intermediate system other than possibly a 
 ml> boss node for a point destination...

i'm wanting to back off of the above because the message's bits tell the
destination system's recipient how the message was originally marked...

 ml> secondly, the intermediate system(s) should ignore those CRAsh, 
 ml> IMMediate and DIRect bits in the message(s) inside the PKTs...

this is what is done in most cases... the key being the attributes of
in-transit netmail are ignored while the local routing and flavor are handled
at the PKT level... the PKT level is different than the packed message level
once intermediate systems are involved... especially since a PKT can contain
more than one packed message and each packed message can have its own set of
attributes...

 ml> this was a problem discovered back in the 80s or early 90s... 
 ml> several tossers and mailers had to be adjusted to deal with this...

 NB> Well, I think Synchronet still handles it that way, where if the 
 NB> CRASH bit is set it will rename the bundle to the CRASH 
 NB> destination's system, and the mailer (binkd doesn't know any 
 NB> better!) will try to send it there.

that should be for locally written messages... in-transit ones not...

 NB> This was an issue with Mystic systems with the "Netmail Crash" 
 NB> option set in the configuration. 

yes, i remember... the problem here also come in with routing... a message may
be marked crash to get it to the next system ASAP... BUT we also have to
remember how crash was handled when BSO first appeared... in that case crash
also meant direct to the destination so no routing was done... dynamic mailer
opened things up and allowed for more options in the handling of netmail...

 NB> I'm not sure if you remember my discussion with Rob about this 
 NB> (although it could have been on IRC, so you probably didn't). If 
 NB> there is any proposals or standards you can point Rob to to fix 
 NB> that issue, that would be great!

i don't recall any specific conversation but i do remember that there was some
discussion...

 NB> Either way, I don't remember Rob doing anything about it, and 
 NB> saying that's how it should be and always has been, etc etc. So I 
 NB> politely asked those Mystic nodes to turn that setting off. :|

that's about all that could be done at that time... it was later that mystic
adjusted its operation in these cases...

)\/(ark