OSEC

Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com
Postfix Archives: Re: Local mail being duplicated

Re: Local mail being duplicated

Andrew McNamara (andrewmconnect.com.au)
Fri, 12 Nov 1999 18:09:30 +1100

>> There are two behaviours currently: the "recursive" default behaviour,
>> and the "forward" behaviour, which is triggered by various things
>> (nexthop not local, primarily).
>>
>> What I'm suggesting is that we make a hybrid - recurse building a list
>> of recipients for forwarding/requeuing (and performing local delivery
>> when it makes sense), but stopping short of any aliases with programs
>> listed on the RHS.
>
>That would not solve the problem, as I am trying to explain ad
>nauseam. With a composite alias "foo: bar, foo", Postfix must
>deliver to both users foo and bar; forwarding would be utterly
>wrong. Either user can expand to a file or command. So your
>strategy would not work.

I still suspect we're talking cross-purposes (either that, or I'm being
completely thick).

In the example you've given, if we expanded foo to two recipients bar
and \foo and forwarded, we wouldn't have a loop (except I think we
might currently just ignore the leading \ - we would have to make it
force local delivery, which is exactly the old sendmail behaviour)?

But... that still leaves aliases with a program on the RHS - we can't
save them as it would be a security nightmare - my suggestion was to
abort the expansion and save the LHS at that point. This means that
program aliases are an exception, but it's fairly mild (and they're
already treated fairly oddly).

Just to put it another way, we expand as much as we can to create a new
"queueable" recipient list, stopping only if it's not safe. Yes, we would
have to include a small amount of magic state in the queue file, but we
should be able to limit it to just the backslash.

For :include:, we expand, and if it has a program alias, we backtrack
and just save it's LHS - we get the current behaviour in this case.
BUT, if there is no program alias, we get the forward behaviour, and no
duplicates.

The same basic deal would apply to .foward as :include:.

Am I still talking complete crap?

 ---
Andrew McNamara (System Architect)

connect.com.au Pty Ltd
Lvl 3, 213 Miller St, North Sydney, NSW 2060, Australia
Phone: +61 2 9409 2117, Fax: +61 2 9409 2111

This archive was generated by hypermail 2.0b3 on Fri Nov 12 1999 - 01:10:34 CST