[der Mouse]

> How is 7 "the _practical_ limit"?  I work with a system that has a
> lot of users with 8-character login names and we've had no trouble.

I've run into problems with them from time to time.  It has to do with
how the login name in certain cases is kept in an 8 byte array, and
not necessarily null-terminated.  DECwindows/Ultrix, at least up to
and including version 4.2, did not allow people with 8 character login
names to log in on the console.  From the syslog messages, it was
clear that it assumed the login name would be null-terminated.

I agree with your other objections, certainly.  I think I came across
more strongly than I intended; I really only meant to make a case for
my opinion that what seems to me to be the main reason for wanting
long login names may not be such a very good reason after all -- and
breaking compatibility with who knows how many systems and application
programs to allow longer login names should, as I said, not be done
without first thinking about whether it's really worth it.

> But I want to be able to change one #define somewhere, rebuild the
> world, and get that limit up to whatever I want (though I would
> accept a hard limit, provided it's reasonably high, like 255).

Now _that_ would be neat.  How many places should it "take", though?
Should rwhod broadcasts have different formats depending on the local
choice for this constant?  Should there be compatibility libraries for
software that assumes the old 8 character array?  What happens when I
run a vendor supplied binary, and they selected a different constant?

> So why is _your_ login - or at least your email address - your
> initials?  What happens when Tanya Irene Hendersson joins the
> company?

No company here -- just me and my wife and a lab full of the kind
machines you need to take apart to move...  (I collect VAXen and
PDPs.)  Point well taken, though.

-tih
-- 
Tom Ivar Helbekkmo
tih@Hamartun.Priv.NO