Date: Wed, 22 Nov 2000 22:55:16 +0100 From: Gerhard Sittig <Gerhard.Sittig@gmx.net> To: opentrax@email.com Cc: billf@mu.org, bugs@freebsd.org Subject: Re: bin/22860: [PATCH] adduser & friends with '$' in usernames Message-ID: <20001122225516.S27042@speedy.gsinet> In-Reply-To: <200011221743.JAA00831@spammie.svbug.com>; from opentrax@email.com on Wed, Nov 22, 2000 at 09:43:42AM -0800 References: <20001122095559.B14080@elvis.mu.org> <200011221743.JAA00831@spammie.svbug.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Nov 22, 2000 at 09:43 -0800, opentrax@email.com wrote:
>
> On 22 Nov, Bill Fumerola wrote:
> > On Wed, Nov 22, 2000 at 07:24:41AM -0800, opentrax@email.com wrote:
> >>
> >> With regards to this bug report I would like to state an
> >> opinion.
> >>
> >> IMO, this patch should be reject for inclusion in FreeBSD
> >> and should be place on a 'freely available shelf'. IMO,
> >> FreeBSD is NOT here to support M$, or their attempts to
> >> obfuscate UNIX. [ ... ]
> >
> > It is really just the way that samba and NT interact. We
> > allow for $ in the username, so why shouldn't our tools allow
> > for it too.
> >
> > This isn't an "exit strategy", this is a bug that adduser and
> > co don't handle it.
Maybe I should rethink the patch and not just allow the '$'
special character at the name's end, but change the patch to
accept any valid username? :> Thinking of the pw(8) part it
should be quite easy to allow for any valid charset for every
single field category. As well as the "valid" checks maybe
shouldn't consist of continued if clauses but could gain
advantage from a switch statement, too.
I don't know where your opinion comes from that this is a
devotion to MS and accepting their intriguing(sp?) UNIX
obfuscation. I'm not even sure whether you're kidding or not,
but I try to take your concern seriously.
I haven't checked the NetBIOS specs (RFC 1001/1002?) for if 0x24
at a netbios name's end is the flag for a domain enabled
workstation. But I guess it is like the other types 0x01, 0x03,
0x1b?, etc. At least I could imagine that's the reasoning where
this convention comes from. And BTW it was IBM to introduce the
SMB network software in the early eighties. MS just has the most
prominent implementations nowadays.
Supporting heterogenous LANs with MS machines in them BTW is a
very strong argument _for_ *BSD deployment -- in any other case
one simply couldn't use it for other things than toying with it
at home or in pure UNIX networks in very rare environments.
> '$' on the end of string is used to indicate end-of-string, in
> AWK, VI, Perl and many more.
You do know the difference between identifiers and regular
expressions, don't you? The end-of-string / end-of-line
placeholder you refer to above is only valid for regular
expressions. Reading "man regex" you could recognize some more
"dangerously dedicated" characters. But avoiding dots, plus' and
minus', dollar signs, parentheses, brackets and the like in
filenames and maybe other identifiers would really shrink your
freedom significantly -- while exactly this freedom is one of the
reasons one uses UNIX for. (so much for the pathetic section :)
To rephrase what Bill F. already stated above: It's just that
the adduser(8) / rmuser(8) / pw(8) tools don't accept already
valid values for the usernames. The PR tries to fix just one of
them and maybe provides a way of making pw(8) easier to extend
for every other case not covered yet. Lower level tools like
vipw(8) already allowed you to create those names and UNIX copes
with them well. So where's the problem?
> IMO, M$'s intent was to obfuscate our tool set, intentionally.
> IMO, thereby, creating a riff between SysAdmins new to UNIX,
> but familiar with Win. Also, most win systems don't advertise
> the fact that '$' is at the end of the line. IMO, further
> obfuscating our system.
You seem to overestimate the fact that Win/DoS users are not as
familiar with the UNIX mechanisms as UNIX users are. :) You
could as well blame UNIX for not telling its users how Windows
would handle the situation they're currently in.
I cannot see the "obfuscation" danger in allowing valid usernames
to be entered at the system admin's tools' frontend. That my PR
was triggered by setting up Samba in a PDC environment is a pure
accident, you can find quite a few threads in the archives where
the question on a regular basis bubbles up why "my usernames are
not accepted although they're valid" (mostly with email setup
environments as a trigger).
> So, IMO, not support this patch directly (via inclusion into
> release material) tells admins that we don't support MS.
> However, IMO, placing the patch in a neutral area, such as I've
> described, still allows them to deal with the issue.
Me and all the other admins are of course dependent on the
goodwill of committers to accept this and any other kind of
extension. :) When a committer has a similar need or immediately
sees the benefit of a proposed patch for a larger user base,
there's usually no problem. If the change is not critical or
even controverse, it just takes a little longer. Until then
everyone of us has the choice of keeping a local CVS repo with
private changes to not wait for committance or MFCs and to have
extensions one doesn't even feel like publishing. And of course
there are patches which never will get accepted since their
disadvantages outnumber the benefits.
And I could quite well live without this extension in the FreeBSD
base. It's just that I felt others wanted to participate, too,
since they could have a similar problem.
> This is my opinion.
Yes, I have mine, too. :) Be assured that I'm one of those not
voluntarily or mindlessly following MS or commercial vendors in
general. Free software and UNIX have had quite an enormous
influence in my life for the last ten years now and I *do* feel
strong about it. Free software has given me freedom and tools I
couldn't have found elsewhere, my education had not been the same
(in terms of flexibility and quality) and I did - and do - gain a
lot from it. That's why I try to give a little back where I
think I can.
PS: Most of the list readers will have experienced this
themselves and can very well skip the pathetic stuff ... :> And
maybe I should f'up to the PR with clearing up that the patch is
just a solution for _one_ example of not accepting valid
usernames by the mentioned tools.
virtually yours 82D1 9B9C 01DC 4FB4 D7B4 61BE 3F49 4F77 72DE DA76
Gerhard Sittig true | mail -s "get gpg key" Gerhard.Sittig@gmx.net
--
If you don't understand or are scared by any of the above
ask your parents or an adult to help you.
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20001122225516.S27042>