Re: [exim] Wishlist item: specify error codes in ACL.

Top Page
Delete this message
Reply to this message
Author: David Woodhouse
Date:  
To: exim-users
Subject: Re: [exim] Wishlist item: specify error codes in ACL.
On Tue, 2004-12-07 at 14:24 +0000, Philip Hazel wrote:
> From the WishList section "things that didn't make it into Exim 4":
>
> . Allow :fail: to specify that 551 be used instead of 550. Maybe allow a 
>   code at the start, optionally? What about :defer:?       


Ah, I hadn't seen that one; my search failed me. Not _quite_ what I was
after, but similar.

> That isn't an ACL, of course. However, "deny" implies 5xx, so you
> shouldn't be able to specify the first digit.


Or there should be a sanity check which prevents you from specifying a
first digit which doesn't match.

> Are you seriously expecting any/some/many/all SMTP clients to actually
> differentiate between 550 and 551?


That was just an example. And no, I don't expect them to differentiate;
but it's a problem of their own making and I don't want to be involved.
If the sender elects to break their own mail in this way I'd rather the
behaviour be _consistent_.

>       But, if a 551 code is used, they [servers] MUST NOT assume that
>       the client will actually update address information or even return
>       that information to the user.


Yeah, I know. Not my problem.

> > I don't actually see RFC1893/RFC2034 ENHANCEDSTATUSCODES on the wishlist
> > either, for that matter.
>
> Hmm. I thought it was a buried item like RFC1892, which has been there
> for over 5 years, but I can't find it either. Not many people have been
> calling for it.


It improves the error reporting when in MSA mode dealing with idiotic
user agents which don't actually display the _text_ in the SMTP
response. Would you believe some MUA authors refuse to display what the
SMTP server actually said on the grounds that it "can't be translated
into the user's language"?

--
dwmw2