|Re: [cross-project-issues-dev] Fwd: Change 73676 insimrel/org.eclipse.simrel.build[master]: Update PTP to RC2|
Contrariwise, surely this is a classic example of why precise feature bounds are bad?
- <features name="org.eclipse.mylyn.github.feature.feature.group" versionRange="[18.104.22.168605250940-rc1]">
+ <features name="org.eclipse.mylyn.github.feature.feature.group" versionRange="[22.214.171.124605252217]">
If the consumption range was perhaps
versionRange="[4.4.0,4.5.0)" or versionRange="[4.4.0,4.4.1)"
the build wouldn't fail every time another project is excessively
tidy. Whether Gerrit traps this problem is pot luck; AFAIAA Gerrit
does not control/observe deletion of P2 repositories.
One of my contributions failed (https://hudson.eclipse.org/simrel/job/simrel.neon.runaggregator.BUILD__CLEAN/491/changes) because BIRT had deleted its RC1 repository. The subsequent build succeeded once BIRT's RC2 repository appeared.
Ensuring that contributions are reasonably stable (last two milestones / release candidates) seems a minor requirement, not suggestion, for SimRel participation.
On 26/05/2016 01:12, David M Williams wrote: