Mailing List Archive

Rich Wales wrote:
>
> I noted today that the e-mail address portion of a user ID is always
> forced to lower case. (g10/keygen.c, function ask_user_id, line 611.)
>
> If this is going to be done, it should really only be done for the
> domain name (the part to the right of the "@" sign). The user name
> (left-hand side) in an e-mail address is explicitly allowed to be
> case-sensitive in the Internet e-mail standard (RFC822, sec. 3.4.7).
>
> As far as I'm aware, very few (if any) modern e-mail systems actually
> care about upper vs. lower case in their own local user names.

Last I checked, login names are CaSe sEnsiTive in *NIX, just like anything
else in UNIX or C. I've never done it (I always use lower case for login
names, and that is accepted practice with UNIX), but you could have one
user named Joe and another named joe. Now-- I telnetted and tried it, and
it seems that some (all recent?) sendmails will kludge the capitalization,
but-- a) I've seen many systems that don't do this. b) There is the case
where alternate capitalizations == different users.

> But
> since the standard permits case sensitivity, it seems to me inappro-
> priate to force the left-hand side of the address to lower case when
> constructing a key's user ID.

Right. The sendmail kludge is to keep the deluge of complaints from
puzzled Windows users to an minimum (I've had to explain this to people
over and over. Even if it's a kludge, I'm personally thankful for it).
Usernames *are* case-sensitive in *NIX and friends, and last I checked most
MTA's run *NIX and friends. Not only is it a standard (And a Good Thing),
I've actually seen more than one address that used mixed case that bounced
when the case was alterred.

Nate

On Sun, Feb 28, 1999, Nathan Kennedy wrote:

> Last I checked, login names are CaSe sEnsiTive in *NIX, just like anything
> else in UNIX or C. I've never done it (I always use lower case for login

Some UNIX systems disallow upper case characters in the username
(Sun OS 4 man page says so...)

> names, and that is accepted practice with UNIX), but you could have one
> user named Joe and another named joe. Now-- I telnetted and tried it, and
> it seems that some (all recent?) sendmails will kludge the capitalization,

It's an option in sendmail (as nearly everything) (F=u for the mailer
makes the name case-sensitive). If you use procmail for delivery, it
will usually force the name to lowercase. So your best bet is
not to use upper case characters (some older termcap entries even assumed
that the terminal has only upper case characters when you login and type
an upper case character first!).

Proposal: make this a compile- (or even run-) time option.

Claus Assmann wrote:

Proposal: make this a compile- (or even run-) time option.

I think that would be overkill. The issue only arises when a user
is creating a new key, or adding a new user ID to an existing key --
something that isn't going to be done very often.

I think it's OK simply to assume the user will know the proper way to
write his/her own e-mail address, and preserve the case (at least of
the left-hand side) as he/she types it.

Rich Wales richw@webcom.com http://www.webcom.com/richw/
See http://www.webcom.com/richw/pgp/ for my PGP key and fingerprint info
*** NOTE: any key generated by me before 1997-09-18 has been REVOKED ***

On 1999-02-27 18:10:41 -0800, Claus Assmann wrote:

> Proposal: make this a compile- (or even run-) time option.

No, just leave in the spelling the user gives us. After all, he
knows best what his e-mail address looks like.

tlr
--
http://home.pages.de/~roessler/

Rich Wales <richw@webcom.com> writes:

> I noted today that the e-mail address portion of a user ID is always
> forced to lower case. (g10/keygen.c, function ask_user_id, line 611.)

You are right (I only had the Postmaster exception in mind). I leave
the case as it is.


Werner

--
ceterum censeo RSA esse delendam

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

* Rich Wales <richw@webcom.com> Fri, 26 Feb 1999
| As far as I'm aware, very few (if any) modern e-mail systems actually
| care about upper vs. lower case in their own local user names.

RFC 822 specifically states that local parts of mailboxes *ARE* case
sensitive, with the singular exception of the Postmaster local part.

| But since the standard permits case sensitivity, it seems to me inappro-
| priate to force the left-hand side of the address to lower case when
| constructing a key's user ID.

Case smashing of the domain part is okay, but case smashing of the local
part is verboten.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v0.9.2 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE22bZUgl+vIlSVSNkRAp7PAKDZ1WSD7t4WAXBQZs0//Thax9DuxwCcDqqz
1j8vW8aRXGCXdZ9Q++76tec=
=oU+o
-----END PGP SIGNATURE-----

--
Rat <ratinox@peorth.gweep.net> \ Ingredients of Happy Fun Ball include an
Minion of Nathan - Nathan says Hi! \ unknown glowing substance which fell to
PGP Key: at a key server near you! \ Earth, presumably from outer space.

I hope people won't feel I'm beating a dead horse, but I feel one of the
issues in this thread requires a bit more clarification.

Earlier, I wrote:

RW> As far as I'm aware, very few (if any) modern e-mail
RW> systems actually care about upper vs. lower case in
RW> their own local user names.

"Stainless Steel Rat" replied:

SSR> RFC 822 specifically states that local parts of
SSR> mailboxes *ARE* case sensitive, with the singular
SSR> exception of the Postmaster local part.

I don't really think we're in serious disagreement here.

My understanding of RFC 822 is that mail transport systems must preserve
upper/lower case distinctions in the local part (left-hand side) of an
address. That is, the local part of an address must be kept the way it
was originally written, and other systems may not change the case of any
of the letters in the local part (though they may, if they wish, change
the case of the letters in the domain name or right-hand side).

RFC 822 does not prohibit local mail delivery systems from ignoring case
distinctions in their own local mailbox names. That is, if a site wants
to treat "fred", "Fred", and "FRED" as the same name, and deliver all
mail addressed to any of these alternatives to a single local user whose
account name is "fred", that doesn't violate RFC 822. (My impression,
FWIW, is that most modern e-mail systems do this very thing.)

But even if a site's local rule is that upper/lower case distinctions
are irrelevant in its own local mailbox names, that does =not= give the
site's software writers permission to change the case on the local parts
of e-mail addresses intended for =other= sites. For all I know, some
other site may treat "fred" and "FRED" as completely different, and
accept mail delivered only to one or the other. Or -- though I can't
imagine any sane sysadmin wanting to do this -- some other site might
have two completely separate accounts named "fred" and "FRED", owned by
two different users. RFC 822 says I'm not allowed to make any assump-
tions in this regard when dealing with other sites' addresses (except
for "postmaster", which I am entitled to assume is always equivalent to
"POSTMASTER", "Postmaster", "pOsTmAsTeR", etc., etc. ad absurdum).

Back in 1982, when RFC 822 was written, there was a tendency by UNIX
mail software writers to force all e-mail addresses to lower case (since
UNIX account names have traditionally used all lower case) -- while, at
the same time, software for TOPS-10, VAX/VMS, and IBM systems tended to
force addresses to upper case (since that was what those systems used).
People needed to be told =not= to change the case of the local parts of
addresses in order to allow e-mail to be exchanged around the net.

When I said that few, if any, modern e-mail systems actually care about
upper vs. lower case, what I meant was that (as far as I'm aware) most
systems today will in fact ignore upper/lower case issues in their own
local mailbox names, and will deliver local mail the same no matter how
the user name is capitalized. Despite this, though, software still
shouldn't munge the case of the left-hand side of an e-mail address --
and that's why I recommended that GnuPG should stop doing this.

Rich Wales richw@webcom.com http://www.webcom.com/richw/
See http://www.webcom.com/richw/pgp/ for my PGP key and fingerprint info
*** NOTE: any key generated by me before 1997-09-18 has been REVOKED ***