Re: [OT] Get users educated, was: Re: Windows and DLLs

[Nils Philippsen <nils wombat dialup fht-esslingen de>]

On Sat, 3 Oct 1998, Jesse D. Sightler wrote:

> -----Original Message-----
> From: Nils Philippsen <nils@wombat.dialup.fht-esslingen.de>
> To: Jesse D. Sightler <jsight@pair.com>
> Cc: Tim Moore <tmoore@tembel.org>; gnome-list@gnome.org
> <gnome-list@gnome.org>
> Date: Saturday, October 03, 1998 6:36 AM
> Subject: [OT] Get users educated, was: Re: Windows and DLLs
> 
> 
> >On Fri, 2 Oct 1998, Jesse D. Sightler wrote:
> >
> >> Yes, but many people do not realize that this is what the somewhat
> confusing
> >> error messages from PRM -Uvh mean.
> >>
> >> These people then often go on to do a RPM -Uvh --force --nodeps to make
> it
> >> work.  If you don't believe that people really do this, then let me
> remind
> >> you that not only have I done it, but several others on the mailing lists
> >> seem to have as well (based on other comments in other threads).
> >
> >If someone doesn't think about what (s)he does -- let her/him suffer. It's
> >the easiest way people will learn to inform themselves about what they do.
> >Be informed or let it be. I don't want to suggest _not_ answering to such
> >people but bring correct answers (not --nodeps --force, but "you have to
> >fetch the package xyz-1.1-4.i386.rpm prior to installing this").
> 
> Yeah, but that isn't the error message received.  The error message is that
> file /usr/lib/gpackge.so.1 and a dozen or so other filenames are all
> required by package foobar.billy.rpm.  And those messages come while
> upgrading gpackage.rpm.
> 
> Doesn't it seem a bit needlessly confusing to give users messages that
> "upgrading" a package will cause problems because apps need the old version?

You're right-- It would be 'nicer' to have "upgrading this would break
things, because packages blah.rpm, blub.rpm and foobar.rpm need the old
version of this package 'gpackagelib-1.0-1'." the current text could be
spit out with --geekmode :-) as command line option.

[snip]
> 
> Ok, so where is it documented that doing an upgrade from one version of an
> RPM to another sometimes requires not doing an upgrade in order to make the
> upgrade not break old package dependences?  Where is it mentioned that the
> long list of possible dependency warnings that the user is confronted in
> mean that they should install two versions rather than upgrading?  I haven't
> seend it.

If this is not already this circumstance belongs into the FAQ.

> 
> >Do you know why so many people can drive a car (I'm speaking for Europe,
> >can't tell for the US -- no offence intended)? Because they have to get a
> >driving licence _before_ doing it. They even pay money for it (mine cost
> >~3100 DM which is about 1900 US$ -- I really did learn a lot in the
> >course). And to "upgrade" from a Volkswagen to a truck they have to
> >"upgrade" their licence, too (realize the similarity to operating systems:
> >Windows <-> car (nice color, air condition, automatic gear switching
> >(don't know the correct term), ...) , UNIX <-> truck (used to get work
> >done, "bad" looks (UI, anyone?), no automatic gear switching (at least
> >over here in Europe), ...)). What I want to say is that people have proper
> >knowledge of what they do in this field, why should it be different with
> >computers? If I use a {coffeemachine, chainsaw, lawnmower, vacuum cleaner}
> >for the first time I inform myself about how to use it (handbook, mum,
> >grandpa, boss, colleague, ...). Few people don't do this and it almost
> >always ends with a catastrophy.
> 
> That's certainly a different philosophy than that of most American's.  :)
> Seriously, most trucks here ARE automatic, and very few people bother
> reading docs (actually reading docs usually makes for a good joke here <g>).
> 

And sometimes they spread boiling hot coffee over their legs and sue
McDonalds for their stupidity. Since this "event" on every cup of coffee
(at least those I drunk in my holidays where I was in Canmada and the USA)
I could read that it is (or may be) very hot. Not that I knew this before.
But at least no one can sue McDonalds again for this because they "should
have read the docs" read: "the should have read what's clearly indicated
on the  outside of the cup". Well, well. There we have this doc thingy
again. I personally believe that people not reading docs should be blown
up in some way (page 1, sentence 1 of the doc: "Do never ever touch the
green button with 'Push me' on it" -- this would make evolution fast
<evilgrin>). My wish would be to get "the best of both worlds" (America,
Europe), but I really don't know, how this should look like or how this
could be achieved.

> >With regard to the car/truck example I don't want to say that people
> >should abstain from using a truck (and it should be horrible to use, and
> >so on), if people improve easy use of a truck, more power to them -- but
> >people driving a truck should always have in mind that it is far heavier
> >than their car "at home". If I use UNIX's "rm" instead of DOS' "del" or
> >Windows Explorer trashcan I must have in mind that deleted files are gone
> >and I should not make mistakes with this command. The same for rpm: It
> >_is_ much more powerful than the Windows Install Wizard procedure, today
> >it _has_ this "archaic" user interface (may change, I think there are a
> >lot of frontends for rpm, people not comfortable with rpm should use
> >them), and using it implies a little more thinking/learning than using an
> >Install Wizard.
> 
> But, once again, this is NOT a problem with the difference between command
> line and GUI.  :)  If they did it with one of the GUI frontends, they would

As I meant to state (I don't know if this was clearly visible) I don't
want to mix up the command line <-> GUI issue with the foolproof <->
professional tool issue. These two issues are completely separable. In
fact, in the last semester I did a GUI for a program that processes
measured data in a specific way. The result had a GUI, but I would sort it
under "professional tool" anyway and not give it to someone who doesn't
know how to use it.

> get the same archaic and confusing messages, and they would once again be
> apt to do a force with the GUI instead of the command line.  Either way, the
> results are equally bad.

IMO this force thing should not be so easily accessible to the users 
(read: only documented in the man page, not in the online help output).
Where it is documented, there should also be the caveats one has to have
in mind when using it (breaking dependencies, read above).

> 
> >
> >>
> >> IOW, this aspect of "upgrading" is needlessly confusing and archaic,
> IMHO.
> >
> >Use a frontend.
> 
> I have never seen a front end that would solve this particular kind of
> problem.  That sorry excuse for an RPM front-end "GLINT" definately
> wouldn't.  :)  Although, I will say that some of the newer GNOME-based
> front-ends for RPM appear to be somewhat promising.

Oh, there are some frontends, never actually tried one of them (modulo
glint) for reasons that might be obvious to the reader :-). You're free to
do one by yourself, though :-).

Bye, Nils
-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Nils Philippsen                  @college: nils@rhlx01.rz.fht-esslingen.de
Vogelsangstrasse 115             @home:    nils@wombat.dialup.fht-esslingen.de
D 70197 Stuttgart                phone:    +49-711-6599405
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Maybe I should patent stupidity so every lawyer will owe me BIG !!
(mpare@cadvision.com)