[comp.mail.sendmail] IBM Do It Again

igb@fulcrum.bt.co.uk (Ian G Batten) (04/09/91)

I've got a sendmail.cf skeleton written using M4 that I use on most
machines here.  My gateway Sun runs IDA, but the others use the same
basic .cf.  Until I put it on an AIX 3.1 IBM RS/6000...

For some odd reason the OD, OF and OH options are thrown out!  This
doesn't surprise me, as the lousy sendmail on AIX 2.something on 6150s
(Hey!  IBM Support!) for some obscure reason didn't call Ruleset 3
before it called Ruleset 0.

Why do these people feel a need to make stupid, incompatible changes to
standard software?  To tie people in, I guess...like the many other
ludicrous changes in AIX in general.  For the sendmail to identify
itself as AIX 3.1/UCB 5.61 is downright nasty --- it ISN'T UCB 5.61
because it DOESN'T support D, F and H.

I don't care is some IBMite tells me its in the documents.  If it claims
to be UCB 5.61 it bloody well ought to be UCB 5.61.

ian

vixie@decwrl.dec.com (Paul A Vixie) (04/11/91)

[Ian G Batten]
>> [...] for some obscure reason didn't call Ruleset 3
>> before it called Ruleset 0.

IDA through 1.2.8 has that property.  S3 is supposed to be idempotent,
so most people just call it from the top of S0 "just to be safe".

>> I don't care is some IBMite tells me its in the documents.  If it claims
>> to be UCB 5.61 it bloody well ought to be UCB 5.61.

Agreed.

Cheers,
--
Paul Vixie
DEC Western Research Lab	<vixie@pa.dec.com>	<paul@vixie.sf.ca.us>
Palo Alto, California, USA	...!decwrl!vixie	...!vixie!paul

rickert@mp.cs.niu.edu (Neil Rickert) (04/11/91)

In article <1991Apr10.214922.23442@pa.dec.com> vixie@decwrl.dec.com (Paul A Vixie) writes:
>[Ian G Batten]
>>> [...] for some obscure reason didn't call Ruleset 3
>>> before it called Ruleset 0.
>
>IDA through 1.2.8 has that property.  S3 is supposed to be idempotent,
>so most people just call it from the top of S0 "just to be safe".

  Just to make sure there is no confusion.

  In test mode, and ONLY in test mode, the IDA sendmails do not first call
ruleset 3.  In normal operations ruleset 3 is always called first.  There
is a message printed out in testmode warning you that ruleset 3 is not
automatically called.  Not calling ruleset 3 in testmode is actually a
of benefit when testing other rules.

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940

ylee@csl.dl.nec.com (Ying-Da Lee) (04/12/91)

In article <1991Apr10.214922.23442@pa.dec.com> vixie@decwrl.dec.com (Paul A Vixie) writes:
>
>S3 is supposed to be idempotent,
>so most people just call it from the top of S0 "just to be safe".

Is it really stated in any "formal" document that S3 is to be
idempotent?  This is news to me and there are a great number of
sendmail.cf's out there that have non-idempotent S3.  None of
Sun's sendmail.cf's I have seen, for example, have idempotent S3.

	Ying-Da Lee
	NEC America
	C&C Software Development Lab
	ylee@csl.dl.nec.com
	uunet!necbsd!ylee

rickert@mp.cs.niu.edu (Neil Rickert) (04/12/91)

In article <1991Apr12.145344.7135@csl.dl.nec.com> ylee@csl.dl.nec.com (Ying-Da Lee) writes:
>
>Is it really stated in any "formal" document that S3 is to be
>idempotent?  This is news to me and there are a great number of
>sendmail.cf's out there that have non-idempotent S3.  None of
>Sun's sendmail.cf's I have seen, for example, have idempotent S3.

 I do not know of anything in sendmail that requires S3 be idempotent.

 It is very common for the first few rules of S3 to convert

   User Name <user@domain>

into just

   user@domain

and it is very common for the last few rules to convert

   user@domain

into

   user<@domain>

 With such a practice, S3 cannot be idempotent.

 What is sometimes done is for another ruleset to convert
'user<@domain>' back to 'user@domain' then run it through ruleset S3 again.
This usually happens after some reformatting, such as converting a!b into
b@a.uucp.

-- 
=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
  Neil W. Rickert, Computer Science               <rickert@cs.niu.edu>
  Northern Illinois Univ.
  DeKalb, IL 60115                                   +1-815-753-6940

ylee@csl.dl.nec.com (Ying-Da Lee) (04/12/91)

In article <1991Apr12.155948.29233@mp.cs.niu.edu> rickert@mp.cs.niu.edu (Neil Rickert) writes:
>In article <1991Apr12.145344.7135@csl.dl.nec.com> ylee@csl.dl.nec.com (Ying-Da Lee) writes:
>>
>>Is it really stated in any "formal" document that S3 is to be
>>idempotent?  This is news to me and there are a great number of
>>sendmail.cf's out there that have non-idempotent S3.  None of
>>Sun's sendmail.cf's I have seen, for example, have idempotent S3.
>
> I do not know of anything in sendmail that requires S3 be idempotent.

That's my point.  It bothers me to see bold assertion that
'S3 is supposed to be idempotent' without supporting evidence.
I have seen no such requirement either in documentation or in
actual practice.

	Ying-Da Lee
	ylee@csl.dl.nec.com
	uunet!necbsd!ylee

vixie@pa.dec.com (Paul Vixie) (04/16/91)

In article <1991Apr12.145344.7135@csl.dl.nec.com> ylee@csl.dl.nec.com said:
# [...]
# idempotent?  This is news to me and there are a great number of
# sendmail.cf's out there that have non-idempotent S3.  None of
# Sun's sendmail.cf's I have seen, for example, have idempotent S3.

No vendor I know of (please look at my return address before you flame me)
ships a sendmail.cf that's worth lining bird cages with.  The closest thing
to a reasonable config is the one that comes in the sendmail kit from UCB.
There are some "after market" configs that work pretty well, including UK
and IDA.  

But nothing in the commercial space works.  Including the one you mention.
-- 
Paul Vixie
DEC Western Research Lab	<vixie@pa.dec.com>	<paul@vixie.sf.ca.us>
Palo Alto, California, USA	...!decwrl!vixie	...!vixie!paul