back to Ubuntu 8.04

[Lonni J Friedman]

On Wed, Jan 14, 2009 at 1:02 PM, C M Reinehr <cmr at amsent.com> wrote:
> On Wed 14 January 2009 13:43, Lonni J Friedman wrote:
>> On Wed, Jan 14, 2009 at 11:27 AM, Michael Hipp <Michael at hipp.com> wrote:
>> > Lonni J Friedman wrote:
>> >> On Wed, Jan 14, 2009 at 10:49 AM, Michael Hipp <Michael at hipp.com> wrote:
>> >>> Lonni J Friedman wrote:
>> >>>> In my experience, Ubuntu's focus has
>> >>>> never been quality.  They push out the shiniest object possible, in an
>> >>>> effort to attract Windows users, and if they happen to get some
>> >>>> quality from upstream Debian, so be it.
>> >>>
>> >>> I doubt there's any real data to establish this (remembering that the
>> >>> plural
>> >>> of anecdote is not data). In my experience I install it, run it, and it
>> >>> works. It doesn't particularly have any more or less issues than any
>> >>> other
>> >>> distro I've ever tried.
>> >>
>> >> I have lots of data.  At $DAYJOB, I install Ubuntu via kickstart
>> >> _hundreds_ of times every month, and then run stress tests.  I do the
>> >> same with OpenSUSE, Fedora & RHEL.  Ubuntu, by far, fails in the
>> >> following areas more frequently than the others (often by an order of
>> >> magnitude):
>> >> 0) kickstart support 100% broken in 8.10.  Bug filed over 2 months
>> >> ago, and remains unfixed to this day.  All requests for a rough
>> >> estimate on when it might be fixed have been completely ignored (and
>> >> I'm not the only person complaining in that Launchpad bug).  Best as
>> >> we can guess is that there is no intent to fix it for 8.10, and it
>> >> might be fixed for 9.04.
>> >> 1) Installer failures.  At random, the OS installer fails to find its
>> >> own packages, and just crashes/dies.  It somehow manages to munge the
>> >> MBR before this, rendering the system unbootable without manual
>> >> intervention.  None of this behavior has ever happened with SUSE or
>> >> Redhat products.
>> >> 2) kernel panics/Oops.  As I mentioned above, I run the same exact
>> >> stress tests on all the distros, using the same exact hardware.
>> >> Ubuntu manages to experience kernel panics/Oops much more often than
>> >> the others.  And there's rarely any clear pattern.  I've seen them all
>> >> over the place, from filesystems, to device drivers, to memory mgmt to
>> >> IO schedulers.  SuSE & Fedora come in second, with RHEL a very very
>> >> distant third (as in just about never).
>> >> 3) random package bugs.  The gcc shipping in 8.10 has a severe flaw,
>> >> against which a Launchpad bug has been opened since well before 8.10
>> >> was released, and remains unfixed to this day.  Fedora10 &
>> >> OpenSUSE-11.1 both ship the same 'version' of gcc, yet both of them
>> >> managed to patch their gcc packages prior to release to fix this flaw.
>> >>  Ubuntu just shipped the brokeness and has ignored numerous complaints
>> >> about it.  There have been many other buggy packages, but this gcc
>> >> issue is yet the latest.
>> >
>> > Interesting stuff. Do you mind if I share it with some Ubuntu-ers that
>> > might be interested? Is your install and stress test duplicatable (sp?)
>> > and would you share it in sufficient detail to allow others to attempt to
>> > duplicate your results? Otherwise, it's still just anecdotes rather than
>> > science and doesn't count.
>>
>> The stress test that I'm referring to (outside of installing the OS
>> repeatedly) is for NVIDIA CUDA QA work, and is therefore not something
>> that is publicly available.
>>
>> However, here are the relevant 8.10 kickstart & gcc bugs that I
>> referenced which are both unfixed for 8.10 (both are seemingly fixed
>> for 9.04 which does no one any good right now):
>> https://bugs.edge.launchpad.net/ubuntu/+source/gccxml/+bug/293807
>> https://bugs.launchpad.net/ubuntu/+source/busybox/+bug/293586
>>
>> > I'll keep your name out of it for now, if you'd prefer.
>> >
>> >> So contrary to popular opinion, I'm not just making this stuff up
>> >> based on an irrational hatred of Ubuntu.
>> >
>> > Ok, fair enough. But your *mannerism* regarding this subject makes you
>> > come off exactly as such. About all you ever bring up is some grievance
>> > from 4 years ago about being ignored by some developers. That's the only
>> > "data" I was aware of.
>>
>> That was the beginning of the miserable experience attempting to work
>> with Ubuntu.  Its just gotten worse from there.
>>
>> >> yet Mark Shuttlesworth
>> >> doesn't seem to care, because his well stated opinion is that their
>> >> competition is Windows, and we all know how the quality is in that
>> >> camp.
>> >
>> > Can you attribute a quote somewhere to this effect? I'd be surprised if
>> > that's really their position since quality is actually the thing that
>> > would cause Windows users to switch IMHO.
>>
>> See this past sunday's NY Times article on Ubuntu where Mark is quoted:
>> http://www.nytimes.com/2009/01/11/business/11ubuntu.html
>
> I saw nothing in the article to justify that statement. I did see: "The
> company's model centers on outpacing Microsoft on both price and features
> aimed at new markets."
>
> As a user of Debian Stable my conclusion is that in their haste to bring out a
> version every six months with the latest & greatest features, they do
> insufficient testing which results in the problems you describe.

That may be part of the problem, however Fedora also has a 6 month
release cycle and isn't similarly afflicted.

>
>> >>> If some newbie (or intermediate, or advanced) Linux user came to me and
>> >>> asked for a distro recommendation I wouldn't hesitate to recommend
>> >>> Ubuntu.
>> >>> Likely they'll be happy, if their needs can at all be met by Linux.
>> >>
>> >> That's unfortunate.
>> >
>> > What? That they'll be happy. ;-)
>>
>> Hopefully they will, but anyone who actually puts any stress of stress
>> on the OS environment will see Ubuntu fall apart around them.
>
> I would never consider Ubuntu for a critical system, but for their target,
> non-technical desktop, market it's probably "good enough".

I suppose, and I'd be willing to accept that if Mark Shuttlesworth
wasn't foisting his LTS (Long Term Support) nonsense on the world.
Talk about putting lipstick on a pig.  Mr. Shuttlesworth has an
inverted concept of what people seek from a long term product.  The
point of providing long term support is that it _starts_ off with a
high degree of quality, and then is supported over time.  Instead from
a quality perspective these LTS releases start off no differently than
any other Ubuntu release, and they are just granting themselves more
time to fix all the defects.

Ubuntu is likely good enough for the Windows crowd.  I have higher standards.


-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
L. Friedman                                    netllama at gmail.com
LlamaLand                       https://netllama.linux-sxs.org