Mailinglist Archive: opensuse-factory (689 mails)

< Previous Next >
Re: [opensuse-factory] Tumbleweed, future directions
  • From: Richard Brown <RBrownCCB@xxxxxxxxxxxx>
  • Date: Sat, 2 Mar 2013 20:44:39 +0000
  • Message-id: <CAA0b23w2iRH2shB+=TBeSX3GgF8GkPYaAePNTbyEy1Aoec3DYg@mail.gmail.com>
On 2 March 2013 18:29, Greg KH <gregkh@xxxxxxxxx> wrote:
Really? Why is this an issue, you don't have to change your repo
pointers anymore, it just takes a long time to do the first upgrade, and
then all is fine.

Not wanting to join in to much with the dead horse flogging, I feel
it's an issue for reasons like the one Lars raises.
So, while functional, wish I could come up for a sensible way to be
smarter about it so users get a more 'gradual' or 'controllable'
rolling experience, while retaining their history.

To illustrate, when users report bugs in Tumbleweed, I fear there is a
tendency to assume it's an issue with Tumbleweed, and therefore bugs
which are also present in the main distribution are left unfixed.

Where have you seen this be reported? I sure haven't.

https://bugzilla.novell.com/buglist.cgi?quicksearch=Tumbleweed

It may only be 9 bugs, but I don't like how a fair percentage of them
(eg 774507, 804238, 770349, 798254, 691629) have languished with
little or no movement. I'm not convinced that all of them (if any) are
true 'Tumbleweed' issues, there doesn't appear to have been any effort
to establish in any of those cited bugs where the fault lies with
Tumbleweed or Release

Maybe I'm reading too much into it, but I get the feeling that,
because it's an add-on product, any bug which mentions it ends up
getting filed (subconsciously or consciously) as a second class
citizen. It's "Oh, Tumbleweed must have broken it" or "I'll wait till
it's filed against Factory/Release". I dont think that does Tumbleweed
any favours, nor the Distribution.
My thought therefore to make Tumbleweed 'factory-based' rather than
'release-based' would address that issue, plus hopefully give us more
Factory testers (which I worry that we're lacking these days) 'through
the back door'

And yes, there are other, probably better, ways to recruit more
testers into Factory, but you were requesting suggestions for future
directions for Tumbleweed, so you got mine :)

What's wrong with the maintenance process? It's quite valuable and
there for a good reason. Why do you want to skip it?

For our released distributions, it's valuable and there for a good
reason - it forces extra pairs of eyes on changes which potentially
could break machines running production workloads.
But it's not without its issues, I have had issues in the past where
I've gone through the process, got the Patch approved, made the patch,
submitted the patch..and months later the patch isn't out yet, even
with poking and prodding. Even when it works right it's an annoying
extra step if you have a fix and you just want it out, but that's a
cost worth paying to avoid breaking users machines.

When it comes to Tumbleweed package selection, am I right in thinking
that new packages don't go through a similar process?
I'm under the impression it's much more a subjective "I think it's
good, I maintain Tumbleweed, and therefore it's in Tumbleweed".
I don't think there is anything wrong with that approach, but
obviously, its got more in common with how we work on Factory than how
we handle updates to our released version.
So I can't help but feel that the Maintenance process is a 'wasted'
extra step in a Tumbleweed environment.
In my mind, if Tumbleweeds selection and review process of a few
maintainers (or just one) is 'good enough', then to me, a logical
option would be to base Tumbleweed on Factory where it follows similar
process.
I feel that Factory is much more stable than it was when Tumbleweed
was created, probably in a big part to stuff like Devel and Staging
projects being used properly to try and keep the really breaky stuff
away from Factory.

Considering this, factoring in the similarities in the selection and
review process, and feeling that I'd rather have Tumbleweed be a
'lead-in' to possibly help test Factory, well, that's how I ended up
with my suggestions, as unwelcome as they may be ;)

Really? When have I ever wanted a newer version in Tumblweed than is in
Factory?

I was certain I'd seen examples - KDE was firmly imprinted in my mind
as a situation where Tumbleweed had been ahead of Factory, so users
lost the latest KDE when Tumbleweed did a release rebase. I can't find
where I got that idea from, so I'm mistaken, I apologise.

Nope, sorry, again, that's way too much work for someone like me doing
this. If you want to try it, please do, and let us know how it goes :)

I doubt I'll have the time to take on something that big any time
soon, but if I do, I'll let you know :)
--
To unsubscribe, e-mail: opensuse-factory+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: opensuse-factory+owner@xxxxxxxxxxxx

< Previous Next >