Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
RE: [dsdp-pmc] The Eclipse Project Dashboard

Hi Bjorn,
 
RE: 1
Just curious, what about the information on http://www.eclipse.org/projects/dev_process/project-status-infrastructure-page4.php , would that still work?
 
RE:2-4
I suppose that your old formulas are probably the best starting point for us to create new formulas.
Can you publish that mentioned documentation page and send us a link?

Cheers,
Martin
--
Martin Oberhuber - WindRiver, Austria
+43(662)457915-85

 


From: dsdp-pmc-bounces@xxxxxxxxxxx [mailto:dsdp-pmc-bounces@xxxxxxxxxxx] On Behalf Of Gaff, Doug
Sent: Monday, May 22, 2006 6:33 PM
To: DSDP PMC list
Subject: FW: [dsdp-pmc] The Eclipse Project Dashboard

Bjorn is not subscribed to the PMC list. Here are his comments.  We should use dash-dev for future comments, I think.

 


From: Bjorn Freeman-Benson [mailto:bjorn.freeman-benson@xxxxxxxxxxx]
Sent: Monday, May 22, 2006 11:31 AM
To: Gaff, Doug
Cc: DSDP PMC list; dash-dev@xxxxxxxxxxx
Subject: Re: [dsdp-pmc] The Eclipse Project Dashboard

 

Martin, Doug, (and the DSDP PMC and dash-dev)
In answer to your email:

1. The project dashboard has been totally changed recently, without any notice to committers. I liked the old dashboard, and I had put a link to it on the dsdp/tm homepage. So my #1 feedback to Bjorn is: if you are planning any changes to publicly visible URLs, let committer know what you're up to. Don't we want to have open processes? I hate broken links like http://www.eclipse.org/projects/dashboard/dashboard_detail.php?project=dsdp.tm that used to work fine.

The project dashboard was a prototype (that's what the big red box warning on the page was about) and was being misused by (among other people) the press. The data was not valid. Rather than continue to tell people that, I just removed it. It was a public beta. I waffle between two positions: providing public betas and then changing them, or not providing any information at all. We look forward to your input via Bugzilla and your help in writing the dashboard code (it's all in the Project Dash CVS).

2. Projects should not define the metrics themselves if they are publicly visible. The dashboard gets unusable if it's not totally clear what's visible. Projects could be enabled to use the DASH databases to make their own computations and publish them on their own homepages. But the common dashboard should work the same for all projects, or it gets totally confusing.

 

3. I agree that different metrics would be useful in different phases of the project, but there should be ONE common definition of what is visible.

I think the best solution is to come up with a community-defined single formula. The original formula was a pseudo-random invention by me. Far better would be to have the community experiment with the raw data and come up with a consensus about the formula or formulas. The old prototype dashboard provided raw data for such experiments, but there was no conversation about the formula other than to say "it's wrong". I'm not sure what to do here.

My next proposal is to allow projects to use project-info.xml to define a formula and then to have dashboard pages that show all the projects "the world as seen by BIRT", "the world as seen by DSDP", etc.

4. Yes, being explicit about the formulas used is important. I don't think that SQL statements are sufficient. There should be some plaintext explanation of what's visible on a report.

There was a whole page about how the old dashboard was computed, and there were links from the dashboard pages, and yet people didn't seem to find it. That page explained the formula, the raw data, everything. The new project-defined or community-defined formulas can do the same.

 

5. Regarding the new metrics:

5a) dsdp.tm project is missing totally.

5b) I liked the metrics on mailinglist, newsgroup and bugzilla activity, I'm missing those. I'm not sure that commits only is a good indicator.

We are working hard to recreate the code that extracts that data. We would be happy to have your assistance. We agree that commits alone is not a good measure.

- Bjorn


Back to the top