On Thu, Jul 27, 2017 at 1:22 AM, Frank Gasdorf <[email protected]> wrote:
> I'm wondering about 'Pure merge commits are excluded' while we stumbled > these days about a validation failure of a merge commit. > I was speaking in the context of the process that harvests data for the IP Log Generator and charts. The ip-validator GitHub hook treats all commits the same way. > > Ist there any know issue that merge-commits on pull-requests do not pass > validations? > https://bugs.eclipse.org/bugs/show_bug.cgi?id=496885 > Is there a different hook between LocationTech projects and others? > Nope. AFAIK, there is only one process that manages this for all three forges (though no PolarSys projects use GitHub). > The origin commit has both, signed off and a signed ECA for committers > E-Mail address > > Can you help here explaining the circumstances? > Every commit is considered individually. The ip-validator treats merge commits the same way that it treats every other commit. If a push contains multiple commits and one of them is a merge commit, then we get a validation failure. If you have more information or suggestions regarding how we can do better, please add your comments to Bug 496885. In the meantime, the ip-validation hook is only able to warn the committer that there is a potential problem. Ultimately, we trust committers to do the right thing: if the validator is getting hung up only on the merge commit, and you're confident that the author has actually signed the ECA for commits that include actual intellectual property, just accept the pull request and move on. If you're not sure and you need me to have a harder look at the pull request, send a note to [email protected]. Wayne > > Thanks in advance > -- > Frank > > > > > 2017-07-26 23:28 GMT+02:00 Wayne Beaton <wayne.beaton@eclipse- > foundation.org>: > >> All commits authored by contributors who are not project committers must >> be signed off. To pass the IP validation hook (i.e. to get a green light), >> the author email must match the signed-off-by email, and we need to have a >> signed ECA for that email address. >> >> The IP tools scan git repositories using a variation of this query: >> >> git log --exclude='_old/*' --branches --no-merges --numstat >> --format=fuller >> >> Basically every commit the repository is indexed regardless of branch. >> Pure merge commits are excluded, but I'm pretty sure that this will include >> the squashed commits (they're just commits, right?). If this query is >> wrong, please open a bug against "Community/IP Log Tool" and we can discuss >> it there. >> >> Note that we support a notion of multiple authors >> <https://www.eclipse.org/projects/handbook/#resources-commit> on a >> commit using "Also-by" tags in the commit message footer. >> >> I can confirm that the index includes commit >> a801ef49386cc13c156ebf2a4dd8eceb9b8c91cf. >> >> The tools only work as well as the data provided. The commit that's >> attributed to you has a different email address than what you've registered >> for your Eclipse Foundation account. It also appears that Jakub is using a >> different email address in Git commits than is registered in his account. >> I've added mappings for these email addresses (that will be included in the >> queries when the indexer script run tonight). >> >> If you can send me a private email with the list of things that you feel >> are incorrect in the log, I can address the issues specifically. I don't >> think that there's any value for the list to see that interaction. >> >> There is more information about IP Logs and the IP Log Generator in the >> handbook <https://www.eclipse.org/projects/handbook/#ip-iplog>. I'll put >> some energy into expanding how the generator decides what to include and >> how to identify and fix problems. >> >> Wayne >> >> >> >> On Wed, Jul 26, 2017 at 1:26 PM, Jens von Pilgrim <[email protected]> >> wrote: >> >>> Thanks for the answers so far. I probably didn't make myself clear. >>> So I will follow up on that: >>> >>> 1) When does a contributor needs to "sign-off" a commit. >>> >>> The ip-validation give a green light if at least one commit was >>> signed-off. >>> E.g. >>> https://github.com/eclipse/n4js/pull/77 >>> >>> The contributor (szarnekow) didn't signed off all commits. We were still >>> able to (squash) merge the commit. >>> >>> My question basically is: What is required (by the ip-validation) to >>> give a green light? >>> >>> 2) How are "Squash and Merge" commits recognized by the ip-tooling? >>> >>> Apparently the ip-tooling is aware of "Squash and Merge" commits. We >>> ONLY use squash-and-merge commits. E.g. >>> >>> https://github.com/eclipse/n4js/commit/a801ef49386cc13c156eb >>> f2a4dd8eceb9b8c91cf >>> >>> In the Who's involved section of the project page, all contributors are >>> listed: https://projects.eclipse.org/projects/technology.n4js/who >>> >>> The can be further analyzed when creating an IP log. The commit above is >>> listed as >>> >>> Brian Thomas Smith (1 contribution) >>> a801ef 70,480 lines 07/21/2017 #85: Update (prepared) download >>> section of N4JS webpage (#86) >>> >>> which is correctly assigned to the contributor. >>> >>> Unfortunately this is where the "everything works fine" seems to stop. >>> In our generated IP log, >>> - I see the "Squash and Merge" commits correctly assigned to the >>> contributors >>> - I see some commits assigned to committers. (to different branches) >>> - I do not see a lot of other commits (from committers) >>> >>> "Any commit that was authored by non-committers are automatically >>> listed. Please see Handling Git Contributions to ensure that Git commits >>> are properly structured." [https://wiki.eclipse.org/PMI/IP_Log_Generator >>> ] >>> >>> The "Handling Git" [https://wiki.eclipse.org/Deve >>> lopment_Resources/Handling_Git_Contributions#Git] doesn't explain our >>> IP log result. >>> At least in our case, a lot of commits from committers are listed as >>> "contributions". >>> >>> What are the rules in the IP-tooling to create the IP-log? What do I >>> need to keep in mind when using "Squash and Merge" commits? Should also >>> committers work on forks instead of branches to keep the IP-log short? >>> Or are there simply bugs in the IP-log generator? >>> >>> Best regards, >>> Jens >>> >>> >>> _______________________________________________ >>> incubation mailing list >>> [email protected] >>> To change your delivery options, retrieve your password, or unsubscribe >>> from this list, visit >>> https://dev.eclipse.org/mailman/listinfo/incubation >>> >> >> >> >> -- >> Wayne Beaton >> Director of Open Source Projects >> The Eclipse Foundation >> >> _______________________________________________ >> incubation mailing list >> [email protected] >> To change your delivery options, retrieve your password, or unsubscribe >> from this list, visit >> https://dev.eclipse.org/mailman/listinfo/incubation >> >> > > _______________________________________________ > incubation mailing list > [email protected] > To change your delivery options, retrieve your password, or unsubscribe > from this list, visit > https://dev.eclipse.org/mailman/listinfo/incubation > > -- Wayne Beaton Director of Open Source Projects The Eclipse Foundation
_______________________________________________ incubation mailing list [email protected] To change your delivery options, retrieve your password, or unsubscribe from this list, visit https://dev.eclipse.org/mailman/listinfo/incubation
