On Wed, 17 Jan 2007, paul.mcilfatrick@??? wrote:
> Was numerical variables not being available outside an individual string
> expansion a design decision in Exim as in Perl I am used to have
> numerical variables available for later use?
I myself am never sure in Perl exactly when they are going to get
reset, but I understand your point. However...
Exim configurations are not programs. If you write, in a router,
something like
local_parts = ^some(regex)(with)(parens)...
domains = whatever
it does not mean that the local_parts regex is used before the domains
condition is tested. (In fact, in this example, domains is checked
first, as documented in 3.12.)
It is true that ACLs are more like programs in that the various verbs
are obeyed in order, but ACLs are latecomers, not arriving till Exim 4.
So the philosophy (and code) was well-established by then.
> Have you ever been asked to modify this in Exim so that these variables
> continue to exist after a string expansion?
There are documented instances when they do - for example, after a
"matches" condition in an Exim filter.
The answer to your question is "no", or more accurately "not that I can
remember (but my memory is not what it was and I may be wrong)".
Looking at the code, I see that in fact $1 is not explicitly unset after
a string expansion. I was mistaken in suggesting this. So why doesn't
your code work? You had this:
discard condition = ${if match{${lc:$mime_filename}} {[.] *(gif|jpg|jpeg|png|bmp)\$}}
!sender_domains = +ok_domains
logwrite = External e-mail from '$sender_address' with invalid attached file type ($1) '$mime_filename' - discard!
The answer is that when Exim is checking +ok_domains, there may be regex
matches, and the numerical variables are also set for some other kinds
of wildcard match. So the resetting of $1 is a consequence of checking
+ok_domains. (This harks back to my comment above about never quite
knowing in Perl when these variables are reset.)
But in general, my original comment stands: You can only rely on $1 etc
remaining available in circumstances where this is documented.
--
Philip Hazel University of Cambridge Computing Service
Get the Exim 4 book: http://www.uit.co.uk/exim-book
