[more details](http://www.cs.unb.ca/~bremner/blog/posts/encoding)
* on the mail receiving end, these messages are either deleted, or ran through
- a filter to be turned into blog posts.
-
- - I'm thinking about how the filter should work. Within a
- mail client, or as a batch tool to scan a mailbox? How to interact with version control, if at all?
+ a filter to be turned into blog posts. As a first step, I have
+[written](http://pivot.cs.unb.ca/git/?p=ikipostal.git;a=blob_plain;f=filters/postal-filer.pl;hb=010357a08e9)
+a filter that decodes the address and writes the message into an appropriate
+mailbox. I would be interested in any ideas people have about security.
* the same plugin can check for comments on a particular page next time the wiki
is generated, and add a link. (more or less done)
> the appropriate mailbox file, with ikiwiki then running to process it.
> --[[Joey]]
>> It is an interesting idea. I like that it uses an arbitrary MUA
->> as a "moderation" interface. One thing it made me think about is
->> how to encode reference (threading) information. One can of
->> course encode this into local-part, but I wonder if it would be
->> better to use header features of mailto (this could also be an
->> alternative to tagged mail addresses for page references).
->> Various client handling of mailto always seemed a bit fragile to
->> me but maybe I am just behind the times. Most headers are ignored, but
->> pseudo-headers in the body might work. For example:
->>[test](mailto:bremner@somewhere.ca?body=X-Iki-Page:%20test%0AX-Iki-thread:%20foobar). I hesitate to use the subject because every mail admin in the
->> world seems to want to add things to the front of it.
->> -- [[DavidBremner]]
+>> as a "moderation" interface. After I killed a debian BTS entry with
+>> clumsy pseudoheader editing I think any
+>> reference info should also be encoded into the address.
>>> Although it is in python, just from reading the Debian ITP, it
>>> looks like
>>> [mnemosyne-blog](http://www.red-bean.com/~decklin/mnemosyne/)