On Fri, 27 Feb 2004, Kelley Reynolds wrote:
> I'm sure this is a far-off wishlist item, but is there any interest at
> all for named variables instead of the acl_ variables?
If people want to take Exim down the road of using more
programming-language like features, then perhaps this should be planned
as a large-scale thing rather than continuing to add more and more
ad-hoc features. I suspect I have gone rather too far already with the
ad-hoccery. Numbered variables are fast to interpret; named variables
would require a whole apparatus to handle the conversion of name to
address. Sure, Exim already has support for binary balanced trees (this
is the "signature" of pretty well all the large-scale programs I've
written) and they might handle this job, but nevertheless a lot of new
apparatus would be needed.
I have a suspicion that the next generation of MTAs are going to be
implemented much more like libraries of functions with "front end
drivers" that are more like programs that configuration files.
No, I am NOT planning on developing such a thing.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
Get the Exim 4 book: http://www.uit.co.uk/exim-book
