Mailinglist Archive: opensuse-buildservice (348 mails)

< Previous Next >
Re: [opensuse-buildservice] Project classifications
  • From: Robert Schweikert <rschweikert@xxxxxxxxxx>
  • Date: Fri, 23 Apr 2010 07:43:32 -0400
  • Message-id: <4BD187E4.1030304@xxxxxxxxxx>


On 04/23/2010 07:36 AM, Adrian Schröter wrote:
Am Freitag, 23. April 2010 13:16:10 schrieb Robert Schweikert:

On 04/23/2010 05:32 AM, Adrian Schröter wrote:
Am Freitag, 23. April 2010 11:12:20 schrieb Stefan Dirsch:
On Fri, Apr 23, 2010 at 10:53:16AM +0200, Adrian Schröter wrote:

Hi,

the openSUSE booster teams has started to improve the

http://software.opensuse.org/search

interface. One important thing is to improve the search
results. It is important from our POV that the project
owner can classify the purpose of his project.

This is currently suggested in this fate request :

https://features.opensuse.org/306232

as

STABLE -- project is considered to be ready to use for End-Users
TESTING -- project should work from point of developer view, but needs
verification
DEVELOPMENT -- project is random state, it might work.
BROKEN -- project is known to be not working atm.


The default value would be "DEVELOPMENT", which is also used
when someone creates a new branch of a package by default.

by default the software search would just show "STABLE"
projects in their results. If no result can be found, the search
will offer to search also for the unstable versions.

Project owners can set this state in their project at any time.

I would like to hear some feedback about the suggested states. Are
they understandable and do they make sense ?

Sounds reasonable to me and I like the idea. I'm just wondering how to
handle
home:<login> projects. They might "win" against an official devel project
in
state DEVELOPMENT since they have been classified as STABLE at some point,
but
are meanwhile in a complete broken state ...

Yes, but a handle for complete unmaintained projects need to be supported
differently.
This mechanism only gives the active developers a chance to flag their
projects.

I'm not sure that the classification on a project level is sufficient.
It appears that one should have the option to make this classification
at the package level.

For the user it is most interesting if he can add the repository or not.
And one single package can make an entire repo broken.

But software search does not search for repositories, it searches for
packages. Those packages happen to be part of a project but a user
searching for couchDB for example is interested in that package and not
in the classification of the project that happens to contain the package.

Projects are a nice abstraction and organization mechanism for
developers and contributers. IMHO from a user point of view in most
cases functionality provided by a given package is of most interest.

Robert

So I would just go for the simple project wide approach for now.

Calculating the trust in single packages, depending on the content and
submitters is a complete different approach (read the trust concept in the
wiki
for this).

bye
adrian


--
Robert Schweikert MAY THE SOURCE BE WITH YOU
Software Engineer Consultant LINUX
rschweikert@xxxxxxxxxx
781-464-8147

Novell
Making IT Work As One
--
To unsubscribe, e-mail: opensuse-buildservice+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: opensuse-buildservice+help@xxxxxxxxxxxx

< Previous Next >