[Top] [All Lists]

Re: environment / Re: Revised drafts posted

2007-03-03 16:14:31


   I have a small question for the environment test: wouldn't it be
simpler to immitate the behaviour with variables:

   If the users want to check for envirnoment property, they include a
well-known file (e.g. system), which imports the relevant information,
which is the same the current environment, and then the sieve-scripts
use it as they want.

Short answer: No.

Longer answer: You're talking about at a minumum supporting some kind of
include mechanism, not to mention the variables extension itself. Include
processing is quite complex and difficult to get right; we have to even agree
on a model for it, and even if we do manage to do that eventually there are
likely to be a variety of naming issues that will make portability an issue.

The main point of environment is to be able to write more portable scripts.
Conflating such checks with an extnesion that by its nature is likely to have
portability issues is a seriously losing proposition.

And as a purely operational matter, even if we implement an include mechanism
(can't say for sure if we would or not - it will depend on the semantics), our
customers are unlikely to enable it given their preference for standalone
scripts stored in LDAP. (This is not my personal preference at all , BTW - I
actually dislike this use of LDAP.)

   The advantage of this simpler approach would be, that the
implementations need not be changed, script-generating utilities
neither, and the sieve language is in fact not extended, but is more


I think doing this with variables and includes is actually quite a bit more
complex and almost certainlyt less portable.


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