Lawrence Greenfield <leg+(_at_)andrew(_dot_)cmu(_dot_)edu> writes:
[12:37]<leg> Why not use LDAP instead of Managesieve? Rant, rant, rant
How would LDAP return a useful error message when the Sieve script
doesn't parse correctly? How would LDAP switch among active scripts?
It seems LDAP would have to be profiled to provide these features,
which effectively creates a new protocol, one that would be more
complicated than Managesieve. Implementation wise, except for
profiling the LDAP implementation, it would also have to interact with
a Sieve parser. It just seems bloated to me, since this LDAP server
isn't likely to be reused for any other purpose, which supposedly was
the point of using LDAP in the first place.
Generally I'm glad to see that there are thoughts about making Sieve a
turing complete language (which I believe it should have been from the
start), with variables and regexps. To cater for big sites that
doesn't want to spend CPU time on behalf of their customers, the
inefficient language constructs could be made optional [1]. No need
to rule out useful language constructs for everyone just because
someone can't implement them.
[1] require ["variables", "while", "backreferences"]