On 24/12/2006, at 2:00 AM, Jason van Zyl wrote:

- build status report (for developers, from Continuum, things that need immediate attention like broken build, failing tests, failing ITs, failing checks)

This report is strictly a record of something that is ready to release. Once all the work was done so that a vote can be presented with it all the pertinent information. The successful docck and verification plugin output is for completeness. Removing the manual tedium.

Yes, I was adding a new report to complete the set. Not something that needs to be done, just trying to be complete.


- development priorities report (for developers, information on what needs attention at any time)

This is captured in other swizzle reports and can definitely be integrated into guides for developers. The plugin report that is generated daily is starting to be an accurate guide for what plugin issues people have:

http://repo1.maven.org/reports/plugins/plugin-issues.txt

Yes, that's what I meant.


- release readiness report (for developers, information on what needs attention before a release can be cut)

They really could use the release report to check themselves.

Yes, we're talking about the same report here. The report John is working on is really a "release readiness" report.


- changes/release report (for users, information on what was in the last release. Could also include announcement text, download links, etc).


These could be integrated into the same report. I think this would be good because then we have one concise report for a release. At the point something is released all this information could be presented. I say this as I've chatted with John and this one piece of information could be attached to a release so that the record of what occurred lives on in perpetuity in the repository. The reporting API is good for outputting html right now but it's not good at aggregating information which is what this would be. It would essentially be a datasource being turned into an XML structure that could be released. Eventually you have an IDE tool that can retrieve those to browse them. Or an Archiva report to create a historical report of releases.

I think it's worth having separate "user" and "developer" reports. The public don't care if docck passed, good documentation should just be a given. I think we are getting at the same thing, though, it's all just different views on the same data.


I think the key differences between 2 and 3 are different handling of JIRA. An issue with 1024 votes needs to be scheduled, but it still shouldn't block a release (eg, it's a feature, and the release in question is only a point release). However, a scheduled issue for the point release will block that release and so needs to be considered.


It wouldn't block the release, it would have already been moved out of the version so that the road map reflected what was being worked on for the release.

I think that's what I was saying too.


I think John is on the right track and that's pretty close to good enough for a first version.

I agree. Was just a bit thrown by the inclusion of "votes" in the issue table, since it's not really relevant once the issue is completed.

Cheers,
Brett


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to