jeff@fluke.UUCP (Jeff Stearns) (09/26/86)
Now that I've installed smail, a nasty question comes up. Should our sendmail
modify "From:" lines?
Background:
We're a uucp site with a dozen or so neighbors. We're running Berkeley
UNIX with sendmail. We just installed smail.
Our standard sendmail rewrites "From" and "From:" lines. Specifically,
it prepends our sitename to the address. This is standard practice here
in uucp-land. All of our uucp neighbors do it, too.
An example. Say we receive this message and relay it to site bletch.
The message comes in through rmail, passes through sendmail, and exits
via uux. It arrives like this:
From foo!bar!joeuser Thu Sep 25 13:29:06 1986
Date: Thu Sep 25 13:29:06 1986
From: foo!bar!joeuser (Joseph User)
To: fluke!bletch!recipient
...
And sendmail emits it like this:
From fluke!foo!bar!joeuser Thu Sep 25 13:29:06 1986
Date: Thu Sep 25 13:29:06 1986
From: fluke!foo!bar!joeuser (Joseph User)
To: bletch!recipient
...
So we tacked on our sitename onto "From" and "From:" as the message passed
through us.
The problem:
I've just installed smail. We're getting domain-ized. We're trying to
make a separation between domain addresses and uucp routes.
What to do with the "From:" line?
I believe that the theoretically correct answer is this:
Don't touch the "From:" line at all. Do your playing with the
"From" line, so that "From" can indicate a valid UUCP path,
leaving the sender's domain address in "From:".
(That last part is a quote from the smail manual page.)
And this is what smail does.
But all of our neighbors tack their sitename onto the "From:" line as well!
There's gonna be lots of unreplyable(?) mail if we start playing by The Right
Rules while everybody else is Doing What The Majority Does. (Recall that
4.2BSD /usr/ucb/mail sends replies to the "From:" address, not the "From"
address.)
What's a mother to do?
--
Jeff Stearns (206) 356-5064
John Fluke Mfg. Co.
P.O. Box C9090 Everett WA 98043
{uw-beaver,decvax!microsof,ucbvax!lbl-csam,allegra,ssc-vax}!fluke!jeffmark@cbosgd.ATT.COM (Mark Horton) (09/28/86)
In article <358@tc-jeff.fluke.UUCP> jeff@fluke.UUCP (Jeff Stearns) writes: >Now that I've installed smail, a nasty question comes up. Should our sendmail >modify "From:" lines? If you have a legal 822 domain From: line, and you modify it by sticking your host name on the front, you violate RFC976 and 822, and your mailer is badly broken. Unfortunately, this applies to most 4.2BSD hosts, and I think the bug is still in 4.3. I think the problem is related to the fact that the From_ and From: lines are tied together by sendmail, and it's very hard to prepend to one without breaking the other. One solution is to install smail, which won't involve sendmail in ordinary pass-through mail. If you have a From: line written with bangs, then there are no standards that apply, and you can just about do what you please with it. But note that it's pointless to stick your host name on the front of a bang path in a From: line to help out a reply command. This doesn't work unless EVERY HOP along the way sticks its name in, and since System III and V don't modify the From: line, the message need go only through one System V hop to render the From: line meaningless. Berkeley long ago gave up on expecting this to work - domains are a much better solution to the reply problem. >But all of our neighbors tack their sitename onto the "From:" line as well! >There's gonna be lots of unreplyable(?) mail if we start playing by The Right >Rules while everybody else is Doing What The Majority Does. (Recall that >4.2BSD /usr/ucb/mail sends replies to the "From:" address, not the "From" >address.) > >What's a mother to do? That mail is generally unreplyable anyway, so you really aren't breaking anything by stopping an unsavory practice. Mark
lyndon@ncc.UUCP (09/30/86)
In article <358@tc-jeff.fluke.UUCP>, jeff@fluke.UUCP (Jeff Stearns) writes: > Now that I've installed smail, a nasty question comes up. Should our sendmail > modify "From:" lines? > > Background: > We're a uucp site with a dozen or so neighbors. We're running Berkeley > UNIX with sendmail. We just installed smail. > > Our standard sendmail rewrites "From" and "From:" lines. Specifically, > it prepends our sitename to the address. This is standard practice here > in uucp-land. All of our uucp neighbors do it, too. > NOT true! (see comments later) > The problem: > I've just installed smail. We're getting domain-ized. We're trying to > make a separation between domain addresses and uucp routes. > > What to do with the "From:" line? > > I believe that the theoretically correct answer is this: > Don't touch the "From:" line at all. Do your playing with the > "From" line, so that "From" can indicate a valid UUCP path, > leaving the sender's domain address in "From:". > (That last part is a quote from the smail manual page.) > > And this is what smail does. > > But all of our neighbors tack their sitename onto the "From:" line as well! > There's gonna be lots of unreplyable(?) mail if we start playing by The Right > Rules while everybody else is Doing What The Majority Does. (Recall that > 4.2BSD /usr/ucb/mail sends replies to the "From:" address, not the "From" > address.) > > What's a mother to do? > -- > Jeff Stearns (206) 356-5064 > John Fluke Mfg. Co. > P.O. Box C9090 Everett WA 98043 > {uw-beaver,decvax!microsof,ucbvax!lbl-csam,allegra,ssc-vax}!fluke!jeff Our configuration is a Convergent MiniFrame running SYS_V with NO sendmail (yes people, there are a *lot* of machines that don't have the foggiest idea about sendmail :-). None of our mailers understands RFC822. All interaction takes place on the "From " headers; The "From:" lines are passed untouched. Strange things begin to happen when I receive mail from site 'a' (4.2BSD) destined for site 'b' (also 4.2BSD). When 'b' receives the mail, the "From " line contains a path of "ncc!a!...", but the "From:" line has "a!...". When b!user tries to reply to the message, it generates a return address from the "From:" line of "a!..." which is obviously incorrect as the mail has to go through "ncc" to get to "a". What to do? There are two options that come to mind. First, "b" can ignore "From:" info, and construct return paths from "From " lines. This maintains compatibility with the dumb *dumb* DUMB Sys_V mailers. It also allows me to continue acting as a mail gateway between 4.2BSD systems. The second option is to use smail/pathalias. When "b" tries to reply directly to "a", the pathalias database will be invoked to find a path to "a". Unfortunately, the default configuration only forces a lookup if the address is of the form "a!b". If the reply was really destined for "a!c!d", the lookup is bypassed. To start messing with the compile time options in smail *will* fix this problem, but strange things start happening as nearly everything gets re-routed. I have thought about building RFC822 into smail (and aliases, and intelligent network gateways, ...). It's about time Sys_V caught up with the rest of the world with respect to mailers. The only problem here is that I keep hearing that "it's coming in the next release on **IX". Arrgh! I would like to hear from other SA's who have run into this problem. The *must* be a reasonable method for dealing with this... -- Lyndon Nerenberg (VE6BBM) {ihnp4,ubc-vision,watmath}!alberta!ncc!lyndon Systems Group - A Div. of Nexus Computing Corp. Envoy_100: Unix
asgard@cpro.UUCP (J.R. Stoner) (10/04/86)
In article <736@ncc.UUCP>, lyndon@ncc.UUCP writes: > > I have thought about building RFC822 into smail (and aliases, and intelligent > network gateways, ...). It's about time Sys_V caught up with the rest of > the world with respect to mailers. The only problem here is that I keep > hearing that "it's coming in the next release on **IX". Arrgh! I have tried to install smail in my SVID validated SYS5r2 system twice. Both times the usual result is some failure to work in the rmail replacement. It would be nice if people who claim to have such a neat interface prepare to write a version that works as advertised on SVID systems as well as the machines used by their elitist friends on VAXen. > > I would like to hear from other SA's who have run into this problem. The > *must* be a reasonable method for dealing with this... > > Lyndon Nerenberg (VE6BBM) {ihnp4,ubc-vision,watmath}!alberta!ncc!lyndon > Systems Group - A Div. of Nexus Computing Corp. Envoy_100: Unix You have heard the word, and the word is BONGA!
diamant@hpfclp.HP.COM (John Diamant) (10/05/86)
In article <358@tc-jeff.fluke.UUCP>, jeff@fluke.UUCP (Jeff Stearns) writes: > Now that I've installed smail, a nasty question comes up. Should our sendmail > modify "From:" lines? > > Background: > We're a uucp site with a dozen or so neighbors. We're running Berkeley > UNIX with sendmail. We just installed smail. > > Our standard sendmail rewrites "From" and "From:" lines. Specifically, > it prepends our sitename to the address. This is standard practice here > in uucp-land. All of our uucp neighbors do it, too. > > The problem: > I've just installed smail. We're getting domain-ized. We're trying to > make a separation between domain addresses and uucp routes. > > What to do with the "From:" line? > > I believe that the theoretically correct answer is this: > Don't touch the "From:" line at all. Do your playing with the > "From" line, so that "From" can indicate a valid UUCP path, > leaving the sender's domain address in "From:". > (That last part is a quote from the smail manual page.) > > And this is what smail does. > > But all of our neighbors tack their sitename onto the "From:" line as well! > There's gonna be lots of unreplyable(?) mail if we start playing by The Right > Rules while everybody else is Doing What The Majority Does. (Recall that > 4.2BSD /usr/ucb/mail sends replies to the "From:" address, not the "From" > address.) > > What's a mother to do? > -- > Jeff Stearns (206) 356-5064 > John Fluke Mfg. Co. > P.O. Box C9090 Everett WA 98043 > {uw-beaver,decvax!microsof,ucbvax!lbl-csam,allegra,ssc-vax}!fluke!jeff Smail is telling you exactly the right thing to do. 4.2BSD Mail does the right thing by using the From: lines since it has sendmail available to it, it is expected to be able to handle 822 style addresses. If it reads an 822 header ("From:") and doesn't understand such a header, it is just plain wrong! SYS V mailers, on the other hand, look at "From " lines, because they don't understand 822 style addresses. "From:" lines will only be generated by sites that understand 822 (or they are wrong). Given that this is the case, if a "From:" line exists, you should assume it is correct. The only time you have to do something funny is if you get are gatewaying a piece of mail from a non-822 site, you will have to add a "From:" line since it doesn't already have one. Here is an example: say your neighbor runs standard SYS V (non-822) and is called "dumbsite", and you are called "gateway.com", and that the user called "user" is sending a message through you. You cannot generate a From: line of user@dumbsite because nobody understands that except dumbsites neighbors. You have a choice of: user%dumbsite@gateway.com or dumbsite!user@getway.com. I have discussed this with Mark Horton (by mail) and he claimed that the second address is O.K. because anyone looking at the From: line must have "@" precedence. While he is technically correct, this is very dangerous, because as you say, there are many sites out there that don't follow rules. The first one should work fine, as long as you make sure your site can translate the address back into UUCP style before transmitting it. You are correct about it being a large problem that people are modifying the From: lines. However, anybody who succeeds in replying to such mail also must have a broken mailer. An "@" is required in a From: line and must take precedence over any other address token (like "!"). So, in fact, tacking on a "!" at the beginning is wrong and is unreplyable by a mailer that conforms to standards. The only thing I can suggest is that the more people who conform to standards, the more likely everyone else will be to conform (because it will hurt them not too). As far as how many people already conform, consider that the entire ARPA Internet does, which means all those From: lines are already not replyable via gateways into the Internet. I encourage you to not modify From: lines and convince your neighbors not to also. The faster we get people to start doing the right thing, the smaller the problem will be -- the real problem is getting a critical mass of people to conform. John Diamant Systems Software Operation UUCP: {hplabs,hpfcla}!hpfclp!diamant Hewlett Packard Co. ARPA/CSNET: diamant%hpfclp@hplabs.HP.COM Fort Collins, CO
ahby@meccts.UUCP (Shane P. McCarron) (10/06/86)
In article <112@cpro.UUCP> asgard@cpro.UUCP (J.R. Stoner) writes: >I have tried to install smail in my SVID validated SYS5r2 system twice. >Both times the usual result is some failure to work in the rmail >replacement. It would be nice if people who claim to have such a neat >interface prepare to write a version that works as advertised on SVID >systems as well as the machines used by their elitist friends on VAXen. I have installed smail on several machines, all of which are running Sys Vr2, and had no trouble at all. A number of local sites have also installed it on everything from Sys III to BSD 4.3, and had no trouble. Could you please send along a little more information about exactly how it is failing? The smail/rmail program is just about too stupid not to work. -- Shane P. McCarron UUCP ihnp4!meccts!ahby MECC Technical Services ATT (612) 481-3589 "They're only monkey boys; We can still crush them here on earth!"
henry@utzoo.UUCP (Henry Spencer) (10/06/86)
Clearly, the way this mess *should* have been handled in the first place
was to recognize that the old (I mean OLD) Unix header, which was a pile
of lines looking, say, like this:
From uucp 22:22 Sept 16
From uucp 22:20 Sept 16 remote from decvax
From uucp 21:11 Sept 16 remote from ucbvax
From joey 09:13 Sept 15 remote from foobar
is in fact a different way of writing:
From: decvax!ucbvax!foobar!joey
Received: here at 22:22 Sept 16
Received: by decvax at 22:20 Sept 16
Received: by ucbvax at 21:11 Sept 16
Received: by foobar at 09:13 Sept 15
or something along those lines. A message which contains "From " lines
should not contain a "From:" line, and vice-versa. Software that talks
between places that speak "From:" (Arpanet; high-level user interfaces)
and places that speak "From " (low-level UUCPNet mail transmission) should
convert. This would eliminate or reduce many, many problems.
Unfortunately, it's far too late for that.
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,decvax,pyramid}!utzoo!henryhokey@plus5.UUCP (Hokey) (10/07/86)
Many sendmail sites *incorrectly* prepend their site name to the From: line. This especially bothersome when the line is in @ format, as the resulting address is *definitely* wrong (user@site becomes relay!user@site, which *must* be parsed as (relay!user)@site by an RFC822-compliant mailer). To a great extent, abolition of this problem would solve the "dumb neighbor" problem mentioned in some of the referenced articles. I won't rehash my belief that *all* RFC822 mail transported in the UUCP domain should be in ! format. Don't forget that even dumb V7 mailers use the To: line if there are multiple recipients. Why not run sendmail on your SysV box? There are versions available which even include nameserver lookup code. -- Hokey
henry@utzoo.UUCP (Henry Spencer) (10/10/86)
> Don't forget that even dumb V7 mailers use the To: line if there are multiple > recipients. Um, not correct: dumb V7 mailers have no concept of mail headers at all, much less any specific notions about the "To:" line. Believe me; we run one. (And as a result we're the only major network hub in this part of the world whose mailer is 100% predictable and 100% reliable.) -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,decvax,pyramid}!utzoo!henry
dan@rna.UUCP (Dan Ts'o) (10/11/86)
In article <7212@utzoo.UUCP> henry@utzoo.UUCP (Henry Spencer) writes: >> Don't forget that even dumb V7 mailers use the To: line if there are multiple >> recipients. > >Um, not correct: dumb V7 mailers have no concept of mail headers at all, >much less any specific notions about the "To:" line. Believe me; we run >one. (And as a result we're the only major network hub in this part of >the world whose mailer is 100% predictable and 100% reliable.) >-- > Henry Spencer @ U of Toronto Zoology > {allegra,ihnp4,decvax,pyramid}!utzoo!henry What happens when you get mail that is misaddressed (wrong host, unknown login name) ? Do you predictably and reliably drop it on the floor ? How can it be 100% when its based on uucp as its transport ? A crashed machine is 100% predictable and reliable, too.
henry@utzoo.UUCP (Henry Spencer) (10/20/86)
> >Um, not correct: dumb V7 mailers have no concept of mail headers at all, > >much less any specific notions about the "To:" line. Believe me; we run > >one. (And as a result we're the only major network hub in this part of > >the world whose mailer is 100% predictable and 100% reliable.) > > What happens when you get mail that is misaddressed (wrong host, unknown > login name) ? Do you predictably and reliably drop it on the floor ? No, we predictably and reliably [we fixed the locking system] forward it to the postmaster [me] for appropriate action. There is no completely satisfactory way of dealing with it automatically. Believe me, there isn't. Even returning it to the sender is an arbitrarily difficult job, because it implies determining the return address from the headers. I think I can safely say that I am smarter than any automated header parser, and *I* sometimes have considerable trouble sorting out the correct return address (no points for getting it wrong, guys, or believing headers that are lying). Sometimes, whether it goes on to its destination or back to its sender is a matter of whose address is easier to determine! > How can it be 100% when its based on uucp as its transport ? The mailer is not based on uucp. The mail system as a whole is. The unreliability of one part of the system (uucp) is not an excuse for unreliability of other parts. > A crashed machine is 100% predictable and reliable, too. Then it's a considerable improvement on sendmail! -- Henry Spencer @ U of Toronto Zoology {allegra,ihnp4,decvax,pyramid}!utzoo!henry