🚀 go-pugleaf

BF> == fenner@postscript.cs.psu.edu (Bill Fenner) writes:
JS> == John Stanley <stanley@skyking.OCE.ORST.EDU>

 BF> If you want to look like part of the Internet, then get a domain name.
 BF> If you don't want to play by the (easy and free) rules, then don't
 BF> complain that the game's no fun.

 JS> What easy and free rule says that UUCP sites have to look like part of
 JS> the Internet?

None.  But if they *do* want to look like part of the Internet, they should
do so, instead of whining about how Internet sites "must" support pathalias
and the .UUCP hack.

 BF> Not usually a problem with the DNS; since the DNS is a distributed
 BF> database, you can be easily in charge of your little section, as
 BF> opposed to having to mail your database updates to central site X.)

 JS> I can be in charge of my "little section" of the DNS? Really? From a
 JS> UUCP site? Just what protocol do YOU know of that will allow a UUCP
 JS> site to change the DNS records for its MX site? 

Email, or telephone.  Or even having an account on your DNS server's
machine so you can make the updates yourself.  Who do you think will be
more responsive, your (single) DNS site, or the map coordinators *and*
everyone everywhere who uses the maps?

 BF> With the UUCP maps, it's easy for automated unpacking systems to
 BF> break, sysadmin doesn't notice, *poof*, you're using old data.

 JS> With the DNS, it is easy to poke incorrect data into the database and
 JS> *poof* EVERYONE is using bogus data. 

And pretty soon you notice that EVERYONE is failing to get you mail, so
someone called your Zone Contact and told them to fix the zone.  If only
30% of the UUCP map people can't get you mail, you may never notice.

 BF> You mean the DNS.  Why not?  That was the original design goal.

 JS> For someone who is adamant that nothing in the UUCP maps could possibly
 JS> apply to Internet sites, why this sudden compulsion that MX records
 JS> must apply to UUCP sites?

Who said MX records "must" apply to UUCP sites?  They can sit out there in
the maps, and maybe I can get mail to them.  They can get in the DNS and
improve their chances.  It's up to them.

 BF> Using the DNS, I have no need to store any information about anyone on
 BF> my computer; I can just look it up.

 JS> And just, pray tell, if you keep no information about anyone in your
 JS> computer, who do you talk to to look it up?

The root name servers are not "about anyone".

 BF> Using the UUCP maps, I have to keep 6.7 megs of map files, and a 1.6
 BF> meg (33k line) paths file.  Which seems like the more desirable
 BF> system?

 JS> Or you keep a <30 character name of a site that will gateway traffic.
 JS> That sounds most desirable.

And *they* keep the maps and path files.  At least until *they* get tired
of doing it, and find another "smarthost".  Who gets sick of routing, and...

Every site that *does routing* with the UUCP maps has to keep them around.
With the DNS, sites can do routing without keeping the entire DNS on their
local disk (or even local net).

 BF> It makes no sense for an Internet-only site to run pathalias; the maps
 BF> are of no use for them.  Therefore, the site must find a way from the
 BF> Internet to UUCP.  The easiest way for it to do that is an MX record.

 JS> To do that reliably requires one MX record for every UUCP site.

It does?  You've never heard of wildcard MX records, I see.

 JS> Passing mail to a gateway requires knowing one address.

Until that gateway changes, goes away, gets out of sync with the maps...

 JS> The latter sure seems to be much simpler than trying to get every UUCP
 JS> in the world to register in a voluntary database.

The UUCP maps are not a "voluntary database"?

 JS> It is, however, less simple that expecting them to.

I don't expect every UUCP site in the world to be in the DNS.  I do expect
those that want to be more reachable from Internet sites to do so.
--
Christopher K. Davis      | ``Usenet seems to run much like the Kif (or,
<ckd@eff.org>   EFF #14   |   for the TV generation, Klingon) high command.
System Administrator, EFF |   Whoever takes action and can be heard wins.''
+1 617 864 0665  [CKD1]   |   --Peter da Silva <peter@ferranti.com>

In article <1gg7eeINNfve@gaia.ucs.orst.edu> stanley@skyking.OCE.ORST.EDU (John Stanley) writes:
>I can be in charge of my "little section" of the DNS?

Yes.

>Really?

You bet!

>From a UUCP site?

Uh-huh.

>Just what protocol do YOU know of that will allow a UUCP site to change the
>DNS records for its MX site? 

mail hostmaster@my.mx.dom.ain

And guess what, if you were to update your UUCP map entry, you'd have to
do something similar anyway.

>>With the UUCP maps, it's easy for
>>automated unpacking systems to break, sysadmin doesn't notice, *poof*,
>>you're using old data.
>
>With the DNS, it is easy to poke incorrect data into the database and
>*poof* EVERYONE is using bogus data. 

And as soon as the Postmaster the MX site sees the problem, he fixes
his DNS database and *poof* everyone is using correct data.

>>|You'll NEVER have the case where EVERYONE is registered in MX.
>>
>>You mean the DNS.  Why not?  That was the original design goal.
>
>For someone who is adamant that nothing in the UUCP maps could possibly
>apply to Internet sites, why this sudden compulsion that MX records
>must apply to UUCP sites?

They don't have to apply.  All Bill, Chris, me, and others are saying if
you want to continue using pathalias, fine, but don't cry when J.
Random Site on the Internet can't send mail to you, since there's no
offical system to provide Internet->UUCP gateways for all Internet members.
Understand?  The solution is to get yourself a registered DNS MX record,
and a site to MX for you.  Plain and simple.

--Dave
-- 
System Administrator, Population Research Institute    barr@pop.psu.edu
	What if there was no such thing as a hypothetical question?

In article <1gg7eeINNfve@gaia.ucs.orst.edu> stanley@skyking.OCE.ORST.EDU (John Stanley) writes:
|In article <Bz4xxE.7M1@cs.psu.edu> fenner@postscript.cs.psu.edu (Bill Fenner) writes:
|>If you want to look like part of the Internet, then get a domain name.  If
|>you don't want to play by the (easy and free) rules, then don't complain
|>that the game's no fun.
|
|What easy and free rule says that UUCP sites have to look like part of
|the Internet?

If you want to send mail to and receive mail from the Internet, then you should
look like part of the Internet, or you should not be surprised when you can't
get mail back from the Internet.  This seems to me to be common sense.

|>It's not the data in the UUCP maps - 3 of
|>the 7 sites that I asked about their entries in d.Top responded.  Two
|>said that the entry was deprecated cruft (one of them said he barely
|>had any UUCP connections at all any more), 
|
|Guess whose responsibility it is for notifying the uucp mapping project
|that a map file is wrong? No, it isn't the uucp mapping project's. It is
|the site who believes his entry in the file is "deprecated cruft".

I didn't suggest that it was the uucp mapping project's fault for not knowing
that the entries were wrong.  I just said that the entries were wrong.  Do
you really think it's a good idea to suggest using a source where nearly
half the data is verifiably wrong?

|>and one said they would route mail if I was a customer.
|
|The same one who will route mail even if you aren't a customer.

This is irrelevant.  Just beacuse they have not installed filters yet doesn't
mean that they won't.  Their policy is, apparently, that they won't route
mail for non-customers, so if you're depending on breaking that policy, don't
be surprised when your mail all disappears one day.

|I can be in charge of my "little section" of the DNS? Really? From a
|UUCP site? Just what protocol do YOU know of that will allow a UUCP
|site to change the DNS records for its MX site? 

mail hostmaster@otc2.psu.edu
Hey, Steve, would you please add an MX record for:
foo.redbeard.com.	IN	MX	20 cs.psu.edu.

Thanks,
  Bill
^D

|With the DNS, it is easy to poke incorrect data into the database and
|*poof* EVERYONE is using bogus data. 

For the part of the DNS that you are responsible for.  Big deal.  I can
put a connection to uunet, with 0 cost, in my map entry, and then rm
/usr/spool/uucp/uunet/* every night.  Which would screw more people up?
For me to say "bogus.redbeard.com. IN MX 20 zing.zang.no.such.host." or
"hogbbs uunet(0)"?

|>You mean the DNS.  Why not?  That was the original design goal.
|
|For someone who is adamant that nothing in the UUCP maps could possibly
|apply to Internet sites, why this sudden compulsion that MX records
|must apply to UUCP sites?

I didn't say anything about MX records applying to UUCP sites.  I said that
the original design goal of the DNS was that everyone would have a domain
name.  I also don't see the leap you took between saying that the UUCP maps
have nothing to do with Internet mail routing and that sites that are
contactable via UUCP should register in the DNS.  Those two statements are
orthogonal, I don't see why you think they're contradictory.

|>Using the DNS, I have no need to store any information about anyone on
|>my computer; I can just look it up.  
|
|And just, pray tell, if you keep no information about anyone in your
|computer, who do you talk to to look it up?

Ever so sorry; I need to store one IP address of a name server.

|Or you keep a <30 character name of a site that will gateway traffic.
|That sounds most desirable.

Sure, except that there's no good way of finding such a site.

|Passing
|mail to a gateway requires knowing one address. The latter sure seems to
|be much simpler than trying to get every UUCP in the world to register
|in a voluntary database. It is, however, less simple that expecting them
|to.

But there's no good way to find such a gateway.  The UUCP maps are no good;
even if that is the intention of the data in d.Top, the data is bogus.

  Bill

In article <1gg7eeINNfve@gaia.ucs.orst.edu> stanley@skyking.OCE.ORST.EDU (John Stanley) writes:

>>If you want to look like part of the Internet, then get a domain name.  If
>>you don't want to play by the (easy and free) rules, then don't complain
>>that the game's no fun.

>What easy and free rule says that UUCP sites have to look like part of
>the Internet?

Rules don't have much to do with it, but if you would like most sites
to be able to reply to your mail you need a domain name.  When mail
is gatewayed among differing systems (uucp, internet, bitnet, etc.)
the return path is often damaged such that a reply is not possible
using the reverse routing information.  If you have a domain name,
the routing doesn't have to be maintained in the message.

>I can be in charge of my "little section" of the DNS? Really? From a
>UUCP site? Just what protocol do YOU know of that will allow a UUCP
>site to change the DNS records for its MX site? 

Wild-card MX records allow this.  All you have to do is maintain one
stable site to act as the gateway and have a forwarder deliver there.
For example, uunet handles MXing and forwarding for the domain *.fb.com,
delivering it to uucp site "afbf05".  As it happens, my mailer uses
fb.com as the name for all our machines and does an alias lookup to
find the right machine for delivery to each user, but if I wanted I
could let addresses go out with subdomains of .fb.com as long as my gateway
machine knows where to send the reply.

>>|You'll NEVER have the case where EVERYONE is registered in MX.
>>You mean the DNS.  Why not?  That was the original design goal.

Personally, I think that unrealistic goals are a sign of a bad design
and more effort should have been put into encapsulating addresses from
incompatible systems so the imbedded routing information could be
used.  But that's kind of irrelevant at this point...


>>It makes no sense for an Internet-only site to run pathalias; the maps are
>>of no use for them.  Therefore, the site must find a way from the Internet
>>to UUCP.  The easiest way for it to do that is an MX record.

>To do that reliably requires one MX record for every UUCP site. Passing
>mail to a gateway requires knowing one address. The latter sure seems to
>be much simpler than trying to get every UUCP in the world to register
>in a voluntary database. It is, however, less simple that expecting them
>to.

It could be really simple: each site that gateways onto the internet could
maintain a wild-card MX so sites using that gateway would automatically
have a return address under the DNS.  A little IDA sendmail header rewriting
magic could even make it happen automatically.  However, sites typically
don't want to lend their domain names to everyone who forwards mail through
them because there are usually other things associated with the name.  And
it's extra work for someone to register another name and set up the MX
so it doesn't get done except within organizations where someone is paid
to do it or in domain parks where a group of users understand the problem
and make the effort to fix it. 

Les Mikesell
  les@chinet.chi.il.us

In article <-ng1H!zgcb@atlantis.psu.edu> barr@pop.psu.edu (David Barr) writes:
|
|They don't have to apply.  All Bill, Chris, me, and others are saying if
|you want to continue using pathalias, fine, but don't cry when J.
|Random Site on the Internet can't send mail to you, since there's no
|offical system to provide Internet->UUCP gateways for all Internet members.
|Understand?  The solution is to get yourself a registered DNS MX record,
|and a site to MX for you.  Plain and simple.


	It seems logical for sites which are Internet/UUCP
	gateways to declare MX records for all the systems
	that they publish in their pathalias map entries -


		sitex.uucp.	MX	10	mail.mysys.mydom.
		sitey.uucp.	MX	10	mail.mysys.mydom.
		...


	I know we don't do this at my workplace, but there
	seems no obvious reason not to do so.


-- 
  ,u,	 Bruce Becker			Toronto, Ontario
a /i/	 Internet: bdb@becker.gts.org	Uucp: ...!web!becker!bdb
 `\o\-e  "...symbolising Excitement and Fun."		
 _< /_		-- Disneyland North investment brochure

In article <1992Dec19.163210.12770@becker.GTS.ORG> uucpadm@becker.gts.org writes:
>	It seems logical for sites which are Internet/UUCP
>	gateways to declare MX records for all the systems
>	that they publish in their pathalias map entries -
>
>
>		sitex.uucp.	MX	10	mail.mysys.mydom.
>		sitey.uucp.	MX	10	mail.mysys.mydom.
>		...
>
>	I know we don't do this at my workplace, but there
>	seems no obvious reason not to do so.

	Most mailers are configured by default to not do DNS lookups on
certain specified domains.  (Most notably .UUCP, .BITNET, and the
now-gone .CSNET)  This is to save undue wear-and tear on the root
nameservers.

	Even if this were not so, with DNS there's no way for one site
to just "declare MX records" for .UUCP and have them be seen by just
anyone.  _One_ site has to be declared as "authoratative" for entire
.UUCP pseudo-domain in order for anything to work consistently.  How is
my mailer going to know what nameserver to look at to resolve the MX
for "becker.UUCP"?

--Dave
-- 
System Administrator, Population Research Institute    barr@pop.psu.edu
	What if there was no such thing as a hypothetical question?