[net.news.b] full names in news 2.7

mark (06/10/82)

With (apparently too) little warning, in 2.7 there is a new feature that
puts your full name in parentheses in the "From" field of your article.
It looks up your entry in /etc/passwd, and checks the gecos field,
which often contains people's full names.  Alas, it often DOESN'T
contain people's full names, or something else comes first, or it's
in some funny format.  Thus, people are sending out articles calling
themselves "pri=-2", "CB2C249", their IBM job numbers, and so on.
(Note that I think the older versions of news - 2.6 and earlier - will
not forward full names since they stop at the first blank.  So you'll
still be getting lots of unsigned articles for some time.  This article
has my name in it but if it passes through any machine that didn't
install 2.7 within 24 hours, to get to your machine, it won't be signed.)

The best solution is for each site to edit their /etc/passwd field
to put the full name FIRST in the gecos field.  Anything after a
comma or semicolon in the field will be ignored.  Thus, you should
list your name in left-to-right order with no commas, e.g.
:Mark R. Horton,foo,bar,mumble: rather than :Horton, Mark,foo,bar,mumble:
or :pri=-2,Mark Horton,foo:  Note that netnews assumes the Berkeley
conventions used by finger(1) - in particular, & will expand into your
login name, thus :& Horton,foo,bar: will expand since my login name
is "mark".  (It will upper case it for you.)  The & feature doesn't help
much if your login is your initials.

If for some reason you can't change this (either you have a piece of
software that can't reasonably be changed that insists on ITS FAVORITE
FIELD being first, or you can't write /etc/passwd, or there are
political problems) the second choice is to change netnews.  The place
to look is the routine gensender in ifuncs.c, and the routine
buildfname that it calls immediately following.  If you fit into this
category I'd like to hear from you what your excellent reason is - if
enough people fall into the same category, maybe some kind of common
standard can be developed that fits everything.

	Mark

davy (06/10/82)

#R:cbosgd:-234400:pur-ee:7300001:000:1868
pur-ee!davy    Jun 10 00:46:00 1982


I can think of numerous reasons *not* to change the password file,
for one thing, ours is 4300 lines long, all in the format

		:LastName, FirstName; Classification:

with possibly some other stuff stuck onto the end of that.  Now, of
course, yours may not be this way.  You may have something like:

		:Class, Priority, Name, Office, Age, IQ, Marital Status:

or any number of other patterns.  Now, how many people have "automated"
account installation (i.e., a program that does it for you)?  We do,
so we'd have to change that, we'd have to change at least two other
programs I can think of also.

Now, frankly, how many of the administrative types are honestly going
to pay some clown to sit down and re-do the whole password file (or
write a program to do it), change all the other programs, re-do any
documentation, etc., all for one program, in particular, netnews?

I seem to remember a lot of people saying that they were having enough
troubles as it is keeping management interested, without things that
cost extra money.

I really don't think that the Berkeley "standard" should be relied on
for something as site-independent as a password file (our password 
file was this way before we even ran Berkeley UNIX).  Instead, the 
routine should be written, with explicit instructions on how to change
it to fit your site.  (I note that Mark has done that to some extent).

I don't mean to sound as critical as I do in this note, however, I
think in the future this will have to be watched for, at least until
some standard *is* written for storage of password files.  I hope I
haven't offended the authors......I mean it as constructive criticism
rather than creative bitching.

--Dave Curry
decvax!pur-ee!davy

P.S. - Just out of curiosity, does the ARPAnet have a standard for such
       a thing?  Is it part of the RFC7533 (or whatever that number is)?

mkg (06/10/82)

I still think that the signature belongs at the end of the article--not
hidden in the From line.  Almost all of the signatures I've seen are
more than the name.  They usually include the english site name, phone #,
and login path.  I favor a .signature file or maybe a line in the .newsrc
file that contains the default signature to be appended to every
article.  Until that time, I will continue to hit pf4 on my HP to say:
   Marsh Gosnell  BTL Piscataway  (201) 981-2758  pyuxbb!mkg

rusty (06/11/82)

Would it be possible to put the full name in the .newsrc file?

mark (06/11/82)

OK, look.  Neither I nor the people at Berkeley had any idea there were
such widely different gecos formats in use.  We just thought that most
passwd files began with a full name.  Clearly this was wrong.

It looks like the answer is going to be to modify netnews to understand
the two most common formats: the Berkeley finger format and the BTL RJE
format.  In addition, those routines will be moved to a separate source
file, so that each site can replace it with whatever is locally appropriate.

Putting your full name in your .newsrc is NOT the answer, although it
may be the best any site can do.  The reason is that it requires each
person to actively make an effort to put his/her full name in the
newsrc file.  The problem is that 99% of the people would not bother.
Relying on a system database which is already there is much better -
it seems like 90% of the systems have the full name somewhere, and
most of them are in the gecos field somewhere.

An alternative I consider acceptable is that the full name go in the
.newsrc file, but the default be something stupid like "Some Guy".
This might be enabled on a per-site basis, or the default might be
taken from gecos for sites allowing it.  Comments are still invited.

	Mark

ARPAVAX:CAD:ESVAX:Cory:cc-18 (06/12/82)

"The GCOS field is used only when communicating with that system,
and in other installations can contain any desired information."

--passwd(5), from V7 Unix Programmer's Manual.

The Berkeley Manual, however, DOES contain (see gecos(5)) an
explicit structure for the GCOS field.  The GCOS field consists
of semicolon-separated strings; the structure is:

	struct gecos {
		char *gc_crn;		/* charge record number	*/
		char *gc_games;		/* games-allowed flag	*/
		char *gc_loglim;	/* login limit in half-hours	*/
		char *gc_spare[7];	/* new fields to be added later	*/
			};

	extern struct gecos *pwgecos();

I think the case against using the GCOS field for net-wide software
is clear.  I suggest, instead, that netnews squirrel away somewhere
a list of real names for people; that enrollnews checks this; and that
the real name appear at the top AND bottom of the article, if possible.

Yes, I know 'enrollnews' doesn't exist yet, but I think it should!
Then all the "how do new users get a .newsrc"/"how does inews find
out what my name is"/"I want to tell everyone on my site not to
talk about the FOOBAR project" issues can be addressed in one place.

Michael Chastain
ucbvax!cory:cc-18      (UUCP)
Cory.cc-18@Berkeley    (ARPA)

sjb (06/12/82)

Why should real names appear at BOTH top and bottom?
I would think that one is enough.

G:asa (06/12/82)

	What about accounts with multiple users?  Seems to me that
automatically tacking one person's name onto an article is somewhat
misleading.
	Also, not to be difficult, but it's sometimes nice to vary your
signature from time to time without some *#%& software crossing you up
(see below).

	Jack The Ripper		ucbvax!G:asa

elman (06/13/82)

We are also in a similar position to the people who asked for complete
2.7 sources: we are running 2.3 netnews and no longer have the 2.6 diffs.

The diffs-style version updates are certaily more economical, but they
can be a problem for those of us who may not keep everything totally
up to date. If you miss one go 'round of diffs, you can really get out
of sync.

Are there others in the same situation? If not, I'll just find someone who
has the 2.6 changes and go from there. If the problem is more widespread,
maybe there's another way of doing things?

	Jeff Elman,  Phonetics Lab, U.C. San Diego
	ucbvax!sdcsvax!phonlab!elman

G:ARPAVAX:CAD:ESVAX:Cory:cc-18 (06/14/82)

I like to see a real name BEFORE I read the article; and I like to
look at it again AFTER I've read it.  The standard business letter has
a 'real name' at the top, and a signature at the bottom.  With netnews,
it's a pain to scroll back to the beginning of the letter!

Michael Chastain
ucbvax!Cory:cc-18      (UUCP)
Cory.cc-18@Berkeley    (ARPA)