Some people get creative† and have UTF8 charactes as DELIMITER_CHARACTERS such as em- or en-dashes. Maybe add those?

† sometimes unintentionally since typing -- can result in a em-dash.

I changed the sender mail address in discourse recently. Now, the mail is very long and clients that break lines at 80 characters will generate a reply like this:

[message body]

/ Greetings

2017-02-05 13:29 GMT+02:00 Very long author name <
[email protected]>:

> [Original Messages is quoted here]
> [...]

The email_reply_trimmer does not strip away the lines between Greetings and the quoted original message.

I've caught another "in reply to"-line in my German Discourse install:

Am 21.04.2016 17:48 schrieb "Max Mustermann"

Here's a (redacted) fragment of the raw email:

Bin gleichzeitig robust, sofern n=C3=B6tig.

Am 21.04.2016 17:48 schrieb "Max Mustermann" <[email protected]=
m
>:

> max.mustermann <https://discourse.pseuco.com/users/max.mustermann> Ma=
x
> Mustermann <https://discourse.pseuco.com/users/max.mustermann>
> 21. April
>
> Vielleicht noch eine wichtige Frage:

would be really awesome for packagers

Some email have the form:

-----Mensaje original----- Enviado: ‎16/‎02/‎2016 14:53

This emails don't are correcting parser

Some signatures aren't delimited by symbols but by specific words or phrases standing alone on a line, such as "Regards", "Cheers", "All the best", "Cordialement", "Bien à vous"...

See this issue for some relevant details.