yum update problems

[Jeff Spaleta]

On Fri, 21 Jan 2005 09:06:37 -0800, Chris Holder <glimmerdark at gmail.com> wrote:
>it could be implied that 'for him, it works like a charm', or
> more appropriately, he hasn't had any issues that have caused him to
> stop using rawhide.

I write to remove all doubt about the implication of the scope of his
comment for those who will find this thread and read into his
statement an impression of wonderous enjoyment for all who use
rawhide.  I insist of people being explicit about their rawhide
experience and to make sure that public statements are not easily
misconstrued. I've seen too many people who aren't prepared to eat
rawhide step into it based SOLELY on seeing someone else go 'works for
me' without the other person providing any caveats about the general
robustness of rawhide from day to day.

Rawhide eats babies. There is no truer statement.

> there's nothing wrong with correcting people on the list, even
> aggressively when needed, but come on.. don't snap at someone because
> they responded, 

I assure you... this was not snapping.  If you want to see an example
of what happens when i snap, i could very easily arrange it to give
you some perspective. For the sake of others on the list however we
can negotiate the  seminar arrangements off-list if you so desire.  I
can tell you that the enrollment fee in the 'watch Jef go off on
someone' seminar course is $3,000 US.

> he's testing rawhide on at least 2 machines, most likely with widely
> varied hardware, and using these boxes as his daily computers.. does
> he frequently post to the list? yeah.. 

And yet even a package maintainer on this same thread before my post
assumed he was using fc3 repos and not development.... which by the
way underscores my point about being as explicit as possible when
making statements about rawhide packages. You just can't assume people
are going to understand your 'implicit' statements.

> get something fixed, he's just passing along what he's seeing, and
> asking a question here and there about it. if all the testers minded
> their own business, and never made a sound.. what exactly would the
> purpose of the list be?

Did i ask him to mind their own business.... no... I told him how to
go about getting the dependancy problems fixed. I actually answered
his question. There are packaging issues that need to be fixed by the
maintainers that prevent the packages from installing with out
dependancy failures. The ONLY way to get that fixed is to get binaries
rebuilt. Now.. he can either download the srpms and edit the spec
files and rebuild them for himself wasting effort that a package
maintainer will have to redo anyways or he can file this issue with
bugzilla and make the maintainers aware. Notice again that one package
maintainer on this thread was already confused by the original post
and though he was talking about fc3 update repos and not rawhide.

This list is meant to be a resource to discuss problems, sure... but
fixes to problems once confirmed are done via bugreports in bugzilla.
This has been and will continue to be how fixes get developed. He
asked how to get this fixed, and I told him. Search bugzilla for
reports, if not reported file a new bugreport.  This list is most
effective when used to ask other testers to confirm problems, with the
end goal of producing a bug report with as much specific information
as possible to aid the package developers and maintainters in fixing
problems. No matter what the problem is however... problems are fixed
primarily via bugreports not primarily with list discussion.

And I stand by my statement... I am greatly concerned about how
someone who has been using rawhide for multiple number of years hasn't
learned that these sort of packaging problems are common and that to
get them resolved is to get bugreports filed that maintainers are
going to see. This would indicate a failure of communication as to
rawhide's goals and how to use rawhide in a responsible manner to
generate useful feedback to package maintainers.

You absolutely can not depend on a maintainer reading a post in a
mailinglist, maintainers are made aware of problems via bugzilla so
that issues can be tracked. He asked how to get it fixed.. i gave him
the the answer. I am very concerned that someone running rawhide
regularly since rhl8 hasn't repeatedly run into several of the
inumerably infinite packaging errors resulting in exactly the same
sorts of dependancy problems. This sort of thing is very common with
rawhide and probably the most trivial sort of problem to confirm.

While it's nice to maybe inform the rest of the testers of the
problems existance with a note like:
"Hey guys FYI i ran into these packaging problems with the latest
rawhide build."

Its much better if you go:
"Hey guys FYI i ran into these packaging problems with the latest
rawhide build and here are the bug numbers that have been filed about
the issues"

Every rawhide user should understand the importance of filing and
citing bug reports about issues and its disheartening to see an
experienced rawhide user fail to exhibit this understanding even when
discussing trivially shallow packaging problems that are historically
common for rawhide.

-jef"did i mention you get a price break if you enroll in 'Jef's guide
to beating a dead horse' seminar as well?"spaleta



> if nothing else, comments like his give users something to find when
> they search through the list archives, trying to find out why
> something is acting the way it does.
> 
> Chris
> 
> --
> fedora-test-list mailing list
> fedora-test-list at redhat.com
> To unsubscribe:
> http://www.redhat.com/mailman/listinfo/fedora-test-list
>