[ https://issues.apache.org/jira/browse/SCM-720?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17962718#comment-17962718 ]
ASF GitHub Bot commented on SCM-720: ------------------------------------ jira-importer opened a new issue, #926: URL: https://github.com/apache/maven-scm/issues/926 **[Fabrizio Giudici](https://issues.apache.org/jira/secure/ViewProfile.jspa?name=fabrizio.giudici)** opened **[SCM-720](https://issues.apache.org/jira/browse/SCM-720?redirect=false)** and commented I've been using for a couple of years a Maven release process where artifacts are first deployed to a local directory and Mercurial/Git changes are not pushed. This makes it possible to roll back all the changes in case of testing, when either the remote SCM repo or Maven repo are not reachable for a transient network error, etc., as well as just retry pushing/deploying the artifacts without re-running the whole release cycle. Final pushing and deploying are performed by re-running Maven with a specified profile. I'm able to perform hg/git push by invoking their executables with the proper parameters by means of the antrun-plugin, but it's cumbersome. It would be so much better to just invoke scm:push. Of course I'm aware that this needs implementation through all the providers. For distributed SCMs, push must be a full push (e.g. Git should push also tags); for non distributed SCMs, such as Subversion, the goals should just do nothing. --- 1 votes, 2 watchers > Add a 'push' goal > ----------------- > > Key: SCM-720 > URL: https://issues.apache.org/jira/browse/SCM-720 > Project: Maven SCM (Moved to GitHub Issues) > Issue Type: Bug > Reporter: Fabrizio Giudici > Priority: Major > > I've been using for a couple of years a Maven release process where artifacts > are first deployed to a local directory and Mercurial/Git changes are not > pushed. This makes it possible to roll back all the changes in case of > testing, when either the remote SCM repo or Maven repo are not reachable for > a transient network error, etc., as well as just retry pushing/deploying the > artifacts without re-running the whole release cycle. > Final pushing and deploying are performed by re-running Maven with a > specified profile. I'm able to perform hg/git push by invoking their > executables with the proper parameters by means of the antrun-plugin, but > it's cumbersome. It would be so much better to just invoke scm:push. > Of course I'm aware that this needs implementation through all the providers. > For distributed SCMs, push must be a full push (e.g. Git should push also > tags); for non distributed SCMs, such as Subversion, the goals should just do > nothing. -- This message was sent by Atlassian Jira (v8.20.10#820010)