[Yum] Re: Yum munging systems?

[Robert G. Brown]

On Fri, 27 May 2005, Axel Thimm wrote:

> On Fri, May 27, 2005 at 09:38:56AM -0400, seth vidal wrote:
> > On Fri, 2005-05-27 at 15:32 +0200, Axel Thimm wrote:
> > > I'm sorry Seth, your choice of language has crossed a barrier. I hope
> > > Karsten will still want to report on his broken alsa systems.
> > > 
> > > On Fri, May 27, 2005 at 09:27:02AM -0400, seth vidal wrote:
> > > > If it sounds like I'm pissed, I am. I'm tired of this horse-shit
> > > > grab-ass of apt vs yum. If you want to use apt, more power to you, if
> > > > you want tell people that yum 'munges' systems then you damn well better
> > > > show me a bug report that backs it up. So far no one has shown me
> > > > anything. So continue recommending apt, feel free, but shut the fuck up
> > > > about yum munging things until you bring me a bug report.
> > 
> > HAHA!
> > 
> > You work in a physics department, right?
> > 
> > I'm shocked you're not exposed to much worse on a daily basis.
> 
> I fail to see the funny part. 4-letter words have no place in any
> technical mailing lists as well as any physics department including
> the one at Duke.

To try to bring a little peace and sanity into the discussion, I think
Seth's point is very clear and was made several times before he lost his
temper.

On a technical mailing list (like this one) the following kinds of
threads are useful and constructive:

  * Bug reports
  * Feature requests
  * Tutorial/HOWTO requests from beginners or experts alike
  * More general discussion concerning the technical future of the
particular tool(s) involved
  * A certain amount of good natured badinage and close-to-topic
discussion.

The following kinds of threads are useless and destructive:

  * Hate reports (unless associated with specific bugs or features, in
which case they are in the first two categories above and ok)
  * Personal invective and ad-hominem attacks
  * Open advertisements for competing toolsets accompanied by
unsupported hate reports
  * Off topic discussions of those competing toolsets or anything else.

Here are some examples to make everything clear:

"Yum sucks and breaks things."  This is BAAAD.  I mean seriously, what
is the point of saying this?  It doesn't help a single person on the
list, either to avoid specific pitfalls or traps as a user or to fix
yum's problems as a developer.  It is in fact polemical, a backhanded
form of ad-hominem attack on the developers, and the kind of discourse
generally excluded from constructive dialogue in ALL fields.

"Yum sucks and breaks things.  Specifically, I used yum to install the
shit-rpm package from Joe's repository with the following yum.conf, and
although it `worked', the dependencies of shit-rpm broke the
critical-rpm package and now my system won't boot."  This is GOOOOD.
Note that even though the comment is still heated, angry and by
implication a wee bit ad-hominem regarding the developers who permitted
all this to happen, it is USEFUL.  Users can beware installing shit-rpm
from Joe's repository in the short run, the developers can worry about
how yum (which is VERY conservatively designed to protect even the
ignorant from being ABLE to break things) managed to break a critical
dependency with this PARTICULAR set of rpms and repositories.

"Apt is a lot better than yum for managing rpms."  This (true or not in
your opinion) is BAAAD.  This list could care less about people's
unsupported opinion on pretty much anything, especially an opinion that
the tool the list is dedicated to isn't as good as some other tool with
its own lists and supporters and good and bad features (none of which
are actually brought into the discussion).  If you think apt is better
and yum thereby pointless, USE apt, and join ITS lists, don't distract
people on this list with your unsupported and thereby worthless
(literally, not as a form of invective) opinion.

"Apt is a lot better than yum for managing rpms.  Apt has the following
features that I really like and that yum is missing.  Also, yum
sometimes munges systems under the following very specific circumstances
(with these particular rpms and repositories) that apt seems to handle."
This is GOOOD.  It isn't just offering an opinion, it is making specific
critical comments that can be used by the yum developers to improve yum,
and by yum users to avoid the circumstances where yum can break things
until the developers have a chance to fix it.  In fact, I recall
discussions very much like this actually occuring on the list.

See?  It's really pretty simple, and has now been said a whole bunch of
times.  Seth finally just lost his temper instead of stating it
patiently yet another time, because it IS (after all) an ad hominem
attack and is openly polemical to make all sorts of extravagent claims
of evil without any sort of specifics or useful accompanying
information.  This is naturally VERY FRUSTRATING to the developers, and
it isn't terribly surprising that it eventually pisses them off royally
and that they take steps to terminate the discussion thread unless/until
it becomes constructive again.

As a last remark, go to the linux kernel list, join up (and be prepared
to be royally buried under a mountain of highly technical traffic:-).
Then post a note that reads something like:

"Linux sucks and inevitably screws up any computer it is installed on to
where the hardware itself won't work even if you later reinstall
something else.  The kernel is clearly badly designed and in many places
is obviously broken.  Microsoft's operating systems are much better."

Nothing more.  No specifics concerning even a SINGLE COMPUTER where
linux broke hardware (or how it happened), which kernel revision was
involved, what distro was being used, how in particular the kernel's
design is "bad" or "broken" or even how Microsoft's are better.

See what kind of language THAT generates, quite possibly from Linus
himself, who can be pretty abrupt in his language if he addresses
something like this at all.  Admittedly, MOST of the linux kernel list
members (probably Linus included) are too busy to deal with polemical
crap like that and would simply cut-and-paste you straight into their
personal .procmailrc whereby you'd instantly become a non-person to them
forever after.  Consequently you'd only get a few hundred pieces of
scathing invective (instead of several thousand), some of it doubtless
pretty obscene, and the list manager would probably add to that a
warning that constructive comment is welcome but that further polemics
would equal list removal and blacklisting at the list server level.

The yum list is actually very, very tolerant of any sort of CONSTRUCTIVE
traffic, including highly critical constructive traffic.  Just keep it
on topic and support assertions with specific examples and avoid the
"yum didn't work for my cousin and in fact broke his entire system but
apt rocks" sort of anecdote, unless you can provide information on just
what your cousin was doing at the time.

Remember, rpm's are by their nature very complex beasts, and are built
by hundreds if not thousands of people with very different programming
styles.  It is quite possible to build in more evil than yum can resolve
into a chain of rpm dependencies, or even to create rpms that openly lie
about things in such a way that later actions can quite possibly break
them no matter what you do to avoid it, although yum does its best and
tries hard to protect users with tell-me-twices and a refusal to
function at all if an installation would require e.g. an rpm --force to
work.  Still, a user CAN use rpm --force by hand, a user CAN mix three
or four internally inconsistent repositories together into a single
yum.conf, a user CAN mess up their system enough that yum "breaks"
things, if you want to blame such a breakage on the tool rather than the
misuse of several tools and incompatible repositories by an ignorant
user.  Even here, the list itself exists to support such a user on the
road to recovery and a more sane usage, and any constructive suggestions
on how to make yum even MORE robust and hence prevent future occurences
(if possible) would doubtless be entirely welcome.

   rgb

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu