[ https://issues.apache.org/jira/browse/BEAM-6613?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Kenneth Knowles updated BEAM-6613: ---------------------------------- Description: Currently the logic is "if RC > 1 then delete the dist.apache.org staging directory". Actually the logic should be "if the staging directory exists and we want to stage a new thing, delete the staging directory". (the overall flow might change but this is surgical) was: Currently, the build_release_candidate.sh re-uses $HOME/build_release_candidate/beam as the git clone. If the RC tag already exists due to a prior build, it crashes. Instead, since that clone failed, it should just not be used the next time. The workflow that makes sense to me is: - Locally tag the intended RC<n> commit - Try to build the RC from that - If the RC build fails, remove tag / discard working branch / etc - If the RC succeeds, push the tag (and if it is on a gradle release plugin commit, those commits) > build_release_candidate.sh fails if RC > 1 but there's no SVN directory on > dist.apache.org > ------------------------------------------------------------------------------------------ > > Key: BEAM-6613 > URL: https://issues.apache.org/jira/browse/BEAM-6613 > Project: Beam > Issue Type: Bug > Components: build-system > Reporter: Kenneth Knowles > Assignee: Kenneth Knowles > Priority: Major > > Currently the logic is "if RC > 1 then delete the dist.apache.org staging > directory". > Actually the logic should be "if the staging directory exists and we want to > stage a new thing, delete the staging directory". > (the overall flow might change but this is surgical) -- This message was sent by Atlassian JIRA (v7.6.3#76005)