[Top] [All Lists]

Re: [sieve] Poll: how to report Sieve runtime errors to the user?

2010-08-24 17:49:20
Clarifying things here:

The server I was talking about was a research prototype that Wolfgang
and I did.  It was meant to eventually turn up in a product, but that
never wound up happening, and some of the decisions we made for
conditions like this were based on the fact that it was a research

We didn't disable the script for things like "fileinto mailbox does
not exist," but syntax errors and other sorts of things got the script
disabled.  We didn't use anything like managesieve ... basically, we
had a special mailbox that users could use IMAP APPEND to stick a
message into, and the content of that message would be used as the
sieve script.  Compilation didn't happen until the script ran, and the
scripts were hand-coded by programmers.

This is obviously not the normal situation.

not people don't even understand that Sieve is what underlies all this, they
don't understand what messages saying stuff like "fileinto used without
first being activated with a require or ihave" mean.)

Indeed.  Telling an end user that her sieve script had an error is
somewhat like asking an end user what she wants the browser to do with
a security certificate problem: it's a WTF? situation.

I really think that Kurt and Ned are saying the same thing here: when
Kurt says that "do an implicit KEEP is enough of an indication that
something's wrong," he's saying what Ned means when he says "I'm not
convinced this is a problem we need to deal with."

I agree with them.  Perhaps if we revise 5228, at some point, we can
change the MUST in the sort of manner that Kurt suggests, but maybe
aim it more at informing an administrator, rather than a user, and add
some vague statement about the method of informing being up to the
implementation, but implementations should keep in mind that it's a
Bad Idea to complain to users about things they won't understand.

sieve mailing list

<Prev in Thread] Current Thread [Next in Thread>