![](https://secure.gravatar.com/avatar/8da339f04438d3fcc438e898cfe73c47.jpg?s=120&d=mm&r=g)
Cameron Simpson writes:
Discourse does not do `In-Reply-To:` very well at all. Here's some headers from the _second_ post in the "Core dev sprint this year" thread:
Message-ID: <topic/17208/60568.898edf234f56cf6f3a661c1a@discuss.python.org> In-Reply-To: <topic/17208@discuss.python.org> References: <topic/17208@discuss.python.org>
I'm tempted to write something uncivil, but instead I'm gonna go hug a puppy and weep.
So at present Discourse's email implementation is buggy. I need to submit a bug report.
Thank you! You may find it useful to cite RFC 5322, section 3.6.4, and emphasize "unique" while mentioning the algorithm for populating References and In-Reply-To presented there.
_However_, someone participating in "email mode" will of course send a message with its own distinct message-id from their source system, and that does not survive the email->discourse->email-out process. [...] I don't expect that to change.
That's just plain obnoxious. Anybody who's in the CCs who participates in "email mode" will get (practically speaking) unfilterable duplicates, and (if there is offline discussion) a bogus new thread. I wonder if this goes all the way through to the backend database (ie, the only id a message gets are its thread id, a timestamp, and some way to ensure a total order in the case of equal timestamps), and the only place in Discourse where the unique Message-ID appears is in the outgoing message. In that case getting any sanity in Discourse email could be very expensive for Discourse. Steve