http://www.efios.com/blog/2004/01/01.html

http://www.rfc.net/rfc2822.html

Thanks, Karl Skibinski!

(See, having my work notes online is a Good Thing.)

E-Mail from Karl Skibinski

Duh, Sacha.

We should store unresolved message-ids somewhere and fill them in later when we see them.

../../notebook/proj/java/org/sacha/socnet/:.java

http://www.w3.org/Protocols/rfc822/rfc822.txt

I can fake it first for testing, anyway, and worry about mboxes later. In fact, I might even end up having formail do the extraction - but that would be cheating now... ;)

SocialNetworkAnalysis

JM Ibanez suggested looking at Apache James and writing a mailet. Nifty idea, if I can get my mind around it!

(I just need something simple. If I get desperate, I'll tokenize the darn thing myself.)

This is where all the magic happens. Starting up the app for every arriving message is probably a bad idea. Better - Java server, then pass it the message.

channel - mailing list

Mailing lists are simpler to track than chat channels. I can work on the References header (and the attribution if necessary).

onMessage: Extract the References header. Find the latest one and root the thread there.

I need a registry of message IDs and authors, then.

Question:

• What about digest users? Should infer based on attribution as well?

../../tmp/piespy/org/jibble/socnet/:.java

http://www.jibble.org/piespy/

I want to run something like this on the PLUG mailing lists. I can reuse their spring model and then build the network based on the References: header.