loginid vs. uid.

[Jon Gefaell]

In article <316 at dynasys.UUCP> jessea at dynasys.UUCP (Jesse W. Asher) writes:
>I have learned to make a distinction between a loginid and the userid.
>The loginid is the actual name of your login.  For example:  jessea.
>The userid is the actual number, your uid.  For example:  110.
>
>I consider these two separate concepts - one is a name and the other a
>number and they are not interchangeable.
>
>My question is does anyone else view them this way, and if not what is
>the relationship?  The reason I make this distinction is because your
>uid can be changed while your loginid doesn't have to be changed - the
>os looks at your uid (the number) to determine who your are in most
>cases.  An example is su changing your uid to 0 (or to whatever your su
>uid is) while leaving your login intact.  Of course you can have both
>changed, but the above leaves me to believe that they are not
>interchangeable concepts.  The uid is not the same as the loginid.
>Does anyone have any comments on this?

 I learned this the hard way when I set up savant and started adding users
I wanted to change the order some people appeared in the /etc/password
file, and their uid's too.. 

Well, this
resulted in some _very_ interesting manifestations. Notably, it seems file
permissions are 'remembered' or whatever in uid form (that is, underneath the
ls output, the file ownership and group ownership must be in uid and gid's...

So, news, who was 101 (right after me, who was 100) I wanted as 100 (so it 
wouldn't interupt a contiguous listing of _real_ people) This meant that
/usr/lib/news and it's files were now owned by 'jon' instead of 'news'
and of course, inews didn't have proper permissions on the history files
(for one thing, but after this became evident I spotted it before any
thing else went wrong (that I noticed)

For those interested, another interesting aspect of this is that there were
no other manifestations of this error within the news system (or any other
that I'm aware of) and that this problem (not being able to post because of
access permission problems on LIBDIR/history.d files) didn't occur until
I ran 'expire' for the first ever 10 days later (waited for a 70M news
fs to start to fill up).

Interesting, at least to me, and illustrates well the diference between
uid gid and their corresponding text tokens. (I believe the later are only 
looked up for their abstract value in user presentation)


As usual, please comment :) I wanna learn....
-- 
+----------- Domain? DOMAIN? We Don't Need No Steeeenkin' Domain! -----------+
| __/\                                                                       |
| \/~~                                                                       |
+-savant!jon at virginia.edu {...}!uunet!virginia!savant!jon jeg7e at virginia.edu-+