Re: Cypherpunks Distributed Remailer (fwd)

From:	IN%"ravage@EINSTEIN.ssz.com"  "Jim Choate" 12-FEB-1997 23:19:40.35

>> From: "E. Allen Smith" <EALLENSMITH@ocelot.rutgers.edu>

>> 	Someone with more experience correct me if I'm wrong (I'm just
>> starting to learn about mailing list management), but isn't saving
>> at least the first bounce from a given address good in order to check
>> what address is the one that's actually bouncing? Admittedly, this only
>> applies to badly-formed bounces. The rest can just have the address
>> noted.

>Hmmm, I can see saving the address causing the bounce in some sort of array
>that is indexed to the majordomo subscription list for a given remailer. It
>would not cost anything but overhead to save the entire message. I guess my
>motivation was not that I cared their mailbox was full or it was a invalid
>address but that I wouldn't be able to deliver at this time. I just don't
>see any reason that it helps me to know why they can't receive mail only that
>they can't.

	Well, if you can't deliver mail because their mailbox is full, then
you should simply wait a few days before trying to deliver mail to that
address. If it's because the address isn't there any more, you should delete
it from the subscription lists. Otherwise, the instant people's mail quotas
overflow (e.g., they're on AOL and they haven't been able to get through the
clogged lines), they get bounced... or you're left with a lot of invalid
addresses. The address causing the bounce is not necessarily determinable
automatically from the bounce message; if it isn't, then you need to keep a
copy around for a human to look at.

[re: bi-directional news-mail gateways]
>> >Want to volunteer?
>> 
>> 	Give me some idea of how much programming knowledge, time, etcetera
>> is required, please, before I can answer. I'd be renting space on some other
>> computer (e.g., an account on cyberpass.net) to do it in; the only higher-level
>> programming language (i.e., beyond Applesoft Basic) that I know anything of
>> is Perl (and I don't know that all that well); and I am somewhat busy with
>> other stuff. In other words, it depends.

>Hmmm, not shure how much programming it would take - never done it myself.
>I think Perl would be an acceptable language. I certainly hope to do the
>majority of the scripts on the CDR in Perl myself. Perhaps there is somebody
>who knows the fine points of how INN cache's messages. Redbeard and I had a
>short discussion about this issue today relating to tracking messages to
>reduce mail-loops. I will discuss it with him and see what happens.

	Given that there are pre-existing news-mail gateways, I'd think that
the programs should be available. My concern on the programming is more in
the realm of keeping the thing up and running.

>I know exactly what you mean. I have two full-time jobs and hadn't really
>thought of taking on the cpunks list in any manner. But hey, that is the
>spice in life...

	Understood; my busy-ness with other stuff comes from being a grad
student.

>Give me a couple of days to look at the situation and if somebody don't beat
>me to it I'll post what I find out. Cool?

	Looks OK. I'll also take a look around for what information I can
find.
	-Allen

>	Well, if you can't deliver mail because their mailbox is full, then
>you should simply wait a few days before trying to deliver mail to that
>address. If it's because the address isn't there any more, you should delete
>it from the subscription lists. Otherwise, the instant people's mail quotas
>overflow (e.g., they're on AOL and they haven't been able to get through the
>clogged lines), they get bounced... or you're left with a lot of invalid
>addresses. The address causing the bounce is not necessarily determinable
>automatically from the bounce message; if it isn't, then you need to keep a
>copy around for a human to look at.

	who should bear the burden of dealing with bounces?
	the list admin or the owner of the bouncing address?

	if you have the time and desire, you can accomodate people
that bounce mail due to a temporarily full mailbox.

	if not, unsubscribe them after N bounces in a 24 hour period.  
it becomes their responsibility not to bounce more than N emails.

	when a list has 1000's of subscribers, i have not have been able
to provide the type of personal service that the first option entails. 
i do make a digest version of every list available.  if the digest bounces,
the bouncer??? bouncee?? stays below the N bounces threshhold 
(for many values of N :).

jmb