[comp.soft-sys.andrew] wpq doesn't seem to see all our users

wollman@EMILY.UVM.EDU (Garrett Wollman) (01/11/91)

[Sorry for all the pos
tings of late...]

I recently did a test installation of WP, to see how well it works in
our environment.  [We run yp, so I do a ypcat passwd >passwd.yp before
building.]  However, it seems that finger does a better job at finding
matching user names than wpq does, unless I'm missing something really
obvious (which is entirely likely, knowing me).

Script started on Thu Jan 10 18:45:33 1991
bash$ f steve
Login name: cavrak                      In real life: Steve Cavrak
Login name: cs294swc                    In real life: Steve W Chappelow
Login name: ee221ss                     In real life: Steve Stratz
Login name: titcomb                     In real life: Titcomb Steve
Login name: steve       (messages off)  In real life: Steve Ackerman
Login name: cs294sfa                    In real life: Steve F Ackerman
Login name: swc                         In real life: Steve Chappelow
Login name: ackerman                    In real life: Steve Ackerman

bash$ wpq -key token +multiple steve
Full name:                 ``Titcomb Steve''
Login name:                ``titcomb''
Source Mask:               ``1''
[. . . fields omitted for privacy . . .]
Surnames phonetically:     ``sdf''
Tokens phonetically:       ``tkmb sdf''

bash$ wpq +multiple steve
Full name:                 ``Steve Ackerman''
Login name:                ``steve''
Source Mask:               ``1''
Surnames phonetically:     ``ekrmm''
Tokens phonetically:       ``sdf ekrmm''
Affiliation abbreviation:  ``evets steve''

bash$ cui
CUI> whois steve
Verifying name list 'steve'...
The name 'steve' is ambiguous.
What did you mean by 'steve'?
    1 - Steve Ackerman (evets steve) <steve+@newton.uvm.edu>
    2 - Titcomb Steve (newton3 titcomb) <titcomb+@newton.uvm.edu>
    3 - None of the Above
script done on Thu Jan 10 18:46:57 1991

I've also noticed that turning AMS_NonAMSDelivery (???) on in AndrewSetup
does not completely eliminate the results of turning on RUN_AMDS_ENV
in site.h; witness the plusses in the display above.  IMHO this is a bug.

Also, even when compiled WITHOUT RESOLVER_ENV, brisk.c still tries to
use the resolver library.  Either brisk should be rewritten *not* to
use the resolver, or it should not be compiled when RESOLVER_ENV is
turned off.  Again, just my opinion (but see below).

-GAWollman

-- 
Garrett A. Wollman - wollman@emily.uvm.edu

Disclaimer:  I'm not even sure this represents *my* opinion, never
mind UVM's, EMBA's, EMBA-CF's, or indeed anyone else's.

Craig_Everhart@TRANSARC.COM (01/15/91)

Finger and WP do rather different sorts of heuristic matching.  Finger
will do what you show: find users by first name.  WP's matching
algorithm was geared more towards mail delivery, and if it finds a match
for userid or surname (or both), it won't match first names.

As the AndrewSetup documentation says, you shouldn't turn on
RUN_AMDS_ENV unless you really expect to be running AMDS in your
environment.  I think, though, that the plus signs you're seeing
(steve+@newton.uvm.edu) would disappear if you turned off both run-time
AMDS (by turning on AMS_NonAMSDelivery) and *also* the
plus-sign-is-valid option (by turning off AMS_UseridPlusWorks in
AndrewSetup).

``brisk'' doesn't quite make resolver calls if RESOLVER_ENV is off, but
it does make reference to the external resolver structure named
``_res''.  I forgot to enclose this in ``#ifdef
RESOLVER_ENV''/``#endif'' pairs, and that should be done.  My regrets.

		Craig