Mailinglist Archive: opensuse-factory (1134 mails)

< Previous Next >
Re: [opensuse-factory] Calling for a new openSUSE development model
On 06/15/2012 01:59 PM, M. Edward (Ed) Borasky wrote:
On Fri, Jun 15, 2012 at 5:44 AM, Jan Engelhardt<jengelh@xxxxxxx> wrote:

In http://lists.opensuse.org/opensuse-factory/2012-06/msg00618.html :

a gcc update would be useful, this probably has to happen very early in
the cycle (right after the release).

It looks like this proposal has potential, but I wonder and ponder.

I went on to compare the development schedules of a few groups. Because
I could not find the freeze dates for older openSUSE anymore (why remove
them from the detailed view?), I took RC1 as freeze point.

#Group Ratio days between merge-phase and frozen-phase
#
openSUSE_11.4 7.50 210/28(RC1)
openSUSE_12.1 8.62 224/26(RC1)
openSUSE_12.2 7.78 210/27(RC1) (prerel-freeze: 6.18, 204/33)
Debian_4 4.73 553/117
Debian_5 2.34 476/203
Debian_6 2.91 536/184
Fedora_17 1.22 111/91
Fedora_16 1.67 105/63
Fedora_15 1.06 104/98

I don't have numbers, but I think Debian has the most packages in a
release, followed by Fedora, followed by openSUSE. The number of
packages matters for estimating effort and hardware resources
required.


And now for the winners:

Linux_3.4 0.24 12/50
Linux_3.3 0.25 15/59
Linux_3.2 0.24 14/58

I don't think it's fair to compare the kernel with an entire distro!


My interpretation: there is not enough testing going on in openSUSE.

Software engineering metrics are a well-established area of study and
have been since the days of Fred Brooks' "Mythical Man-Month". These
are wheels that need not be re-invented. I think the first order of
business in the post-12.2 planning is to come up with goals for the
next release relative to the other community distros and the user
communities.

1. Given that a *new* user has decided to try Linux, how does he/she
decide which distro?
2. What makes a long-standing Linux user switch distros?
3. How does openSUSE propose to attract new users faster than Fedora,
Ubuntu and Mint?

This, as you state requires planing, which is what we have avoided/ignored so far. To this point our releases have mostly grown organically and this has put us into hot water w.r.t. the time based release schedule every now and then, more so now than previously.

Following the "planning" argument would lead to another discussion thread.

"Do we as a community want a feature planned release?"

There are a lot of implications that I do not want to get into and I don't really want anyone to answer or respond to this in THIS thread as it is only tangentially related to the development model discussion.

For me the model is more about how packages get to Factory, how we deal with SRs etc. Even if we planned features, if we didn't change the dev model the flow of packages to Factory would be the same, thus my claim the the planned "vs." organic is only tangentially related.

Later,
Robert


--
Robert Schweikert MAY THE SOURCE BE WITH YOU
SUSE-IBM Software Integration Center LINUX
Tech Lead
rjschwei@xxxxxxxx
rschweik@xxxxxxxxxx
781-464-8147
--
To unsubscribe, e-mail: opensuse-factory+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: opensuse-factory+owner@xxxxxxxxxxxx

< Previous Next >
This Thread