Closed FND closed 9 years ago
The same bug occurs for question marks. Before working on this I think we should agree about valid characters within a mention.
Accidentally I introduced the bug here: 47b49868476141daf0b3ca6a0cab3668198720c3 Thanks to @pschirmacher for pointing this out.
Maybe we should just change the 2nd capture group back to \w+ again.
Maybe we should just change the 2nd capture group back to \w+ again.
Yup, that sounds good.
Accidentally I introduced the bug here: 47b4986
IMHO we should not linkify everything that has an @
in front but instead have a list of valid usernames. Just my 2ct
Was thinking about the same as @aheusingfeld. If this becomes common sense I would start implementing this.
I agree, but we currently don’t have a list of valid user names as we don’t access any user directory (and I would be extremely reluctant to introduce that dependency).
I agree with @stilkov: we should just do the Simplest Thing and assume usernames are [0-9a-zA-Z]+
assume usernames are [0-9a-zA-Z]+
Why not just [0-9a-z]+
? Anyway I didn't ever use the linkified mentioned usernames and as the worst that could probably happen is linking to a list for a user that doesn't exist...........
Why not just [0-9a-z]+?
I think @FND would disapprove.
I think @FND would disapprove.
I know ;)
the worst that could probably happen is linking to a list for a user that doesn't exist
Exactly - verification might happen in whatever system processes mentions (e.g. sending notification e-mails, there checking against an index of known users/addresses makes sense).
a message like ...
... leads to the following HTML being rendered:
that should read