On 4 July 2013 13:14, sebb <[email protected]> wrote: > On 4 July 2013 12:52, Stephen Connolly <[email protected]> > wrote: > > On 4 July 2013 12:32, sebb <[email protected]> wrote: > > > >> On 4 July 2013 11:05, Stephen Connolly <[email protected] > > > >> wrote: > >> > I have asked the legal-discuss list for an opinion on test data sets > and > >> > license headers. From my reading of the current ASF position: > >> > http://www.apache.org/legal/src-headers.html#faq-exceptions we do not > >> > currently have an exception for test data sets. > >> > >> I undertstand that some test data files cannot have AL headers. > >> > >> However, surely any exception which might be agreed does not cover > >> unit test cases in Java files? > >> Nor the poms needed to run the tests. > >> There are several such in the source archive. > >> > > > > Are these in src/test/resources/ or in src/it/ because if so they are > test > > data not software we distribute and build. > > One example is: > > > maven-core/src/test/projects/lifecycle-executor/project-with-additional-lifecycle-elements/src/main/java/org/apache/maven/lifecycle/test/App.java >
They are all in the rat exclusions list as test resources. They are test data. They are not part of Maven core or part of the Maven plugins. > > They are all distributed, because they are in the source release. > > > Without inspecting each and every test data file and the corresponding > test > > case (which invokes the test case using e.g. maven verifier or maven > > invoker... i.e. files in src/test/java which clearly must have a license > > header) we cannot be certain that adding a license header will not affect > > the results of the test... yes the vast majority of them can probably > have > > the header added successfully... but I do not see anyone standing up > > willing to undertake that mammoth analysis task. > > Why not just add the header and see what breaks? > Aha! you miss the point. These are all tests that pass *because* the bug they fix is fixed. If we add the header, how do you know that adding the header does not render the test useless. If there is a regression the test is supposed to fail. If we add the header and introduce a regression and the test does not fail *because* of adding the header then we have rendered the test useless. Thus we cannot "just see what breaks" because the tests are all passing right now. Only by re-introducing the bug that the test was designed to detect can we confirm that the presence of the header does not affect the strength of the test. The only way to be sure that adding the header does not affect the test sensitivity is to inspect each test by hand, evaluate what it requires of the test data and then see if adding the header will affect sensitivity. > > I would much rather get an exception that lets us release and then turn > the > > screw and tighten which files need the exception to remain and which can > > have the headers added as we progress through multiple releases. That is > > the approach which will help the community as us taking what could be > > *years* to get the test cases with headers in *every* place where they > can > > safely be put quite frankly would kill this project. > > > > > >> > Pending the outcome of that discussion I will have to be > >> > > >> > -1 > >> > > >> > If the outcome is that we do not need to do anything for test data > sets, > >> > then I would be happy to switch to +1. > >> > > >> > If the outcome is that we need to add some additional text to the > NOTICE > >> > files to cover the test data sets, then we will need to respin as > nobody > >> on > >> > the PMC can vote +1 if we are aware that the release is in violation > of > >> the > >> > ASF policies and we would be neglecting our governance role. > >> > > >> > If the outcome is that we need to add the license headers to all the > test > >> > data files, then I think the PMC will have to review what we want to > do > >> as > >> > adding license headers to every file in the test data set runs the > risk > >> of > >> > invalidating the test data and that is an unnecessary risk that would > >> > cripple the project and as such I would be looking for the ASF to > change > >> > such a decision and provide us with a means of using the NOTICE file > to > >> > cover the test data. > >> > > >> > I hate being petty, but unfortunately that is part of the governance > role > >> > that the PMC is tasked with... :-( > >> > >> I would not classify it as unfortunate. > >> > >> ASF committership and PMC membership are both privileges, not rights, > >> and come with certain responsibilities. > >> In particular to ensure that releases are available under the Apache > >> License, and don't contain any suprises for the end users. > >> > > > > I started the drive to get the licensing tidied up before you started > > putting your attentions here: > > > https://github.com/apache/maven/commit/bfcf03d42c37bce5bf16a1ab689073109d191bf7 > > > > I understand why the PMC has to do this... > > > > Quite frankly it was only when I groked what it means to be a member of > the > > ASF that I got the fuller understanding of the role of the PMC that I > have > > now. > > > > I think there are a lot of people in the community that do not understand > > why the ASF does things the way it does. > > In that case they need to be helped to understand. > IMO we could do a lot better at documenting the rationale for the > procedures that have been developed. > > > Such people are prone to see the PMC being pernicious as "meddling" and > > "getting in the way"... and may even see my -1 vote as "unfortunate". I > was > > just framing my actions so that they are not confused. > > Well using "unfortunate" confused me - I thought you were saying that > the role was not necessary or was due to some arbitrary rules. > > > Aside: If a committer does not like that we (the PMC) have to be this > way, > > and does not think they could uphold the responsibility to conduct their > > reviews diligently taking care of the legal responsibilities that the > > ASF tasks the PMC with then I would recommend that such a committer > reject > > a nomination to the PMC if we were to offer it... in some ways being on > the > > PMC is a thankless job... but it is one that I agreed to do... so until I > > get fed up, so be it! > > Apache committers also have responsibilities, even though their votes > do not formally count. > In particular, committers are responsible for ensuring that code they > check in to the CMS has the appropriate license. > > > -Stephen > > > > > >> > - Stephen > >> > > >> > > >> > On 1 July 2013 03:56, Barrie Treloar <[email protected]> wrote: > >> > > >> >> On 1 July 2013 06:52, sebb <[email protected]> wrote: > >> >> > Another problem: the NOTICE file contains the following spurious > text: > >> >> > > >> >> > > >> >> > >> > ========================================================================= > >> >> > == NOTICE file corresponding to the section 4 d of > >> >> == > >> >> > == the Apache License, Version 2.0, > >> >> == > >> >> > == in this case for the Apache Maven distribution. > >> >> == > >> >> > > >> >> > >> > ========================================================================= > >> >> > > >> >> > This must not be present in NOTICE files, which are required to be > as > >> >> > short as possible (but no shorter). > >> >> > > >> >> > ASF NOTICE files must start as per the following example: > >> >> > > >> >> > ==================== > >> >> > Apache Maven > >> >> > Copyright 2001-2013 The Apache Software Foundation > >> >> > > >> >> > This product includes software developed at > >> >> > The Apache Software Foundation (http://www.apache.org/). > >> >> > ================== > >> >> > > >> >> > Note also that the phrase should be "developed at" not "developed > by" > >> >> > - the distinction is important. > >> >> > > >> >> > Furthermore, the NOTICE file refers to additonal 3rd party > software, > >> >> > but there don't appear to be any LICENSE files for the software. > >> >> > The licenses should either be in LICENSE.txt or linked therefrom. > >> >> > >> >> From what I can tell we have been failing this since August 2010. > >> >> > >> >> > >> > https://git-wip-us.apache.org/repos/asf?p=maven.git;a=blob;f=apache-maven/NOTICE.txt;hb=bfaf11090a212f8445f2ad929af8acce5a984bf0 > >> >> > >> >> I can't find change history for > >> >> http://www.apache.org/legal/src-headers.html so I don't know if we > >> >> have been failing all the time, or since it was changed. > >> >> > >> >> And I can see you've raised http://jira.codehaus.org/browse/MNG-5487 > >> >> to track this. > >> >> > >> >> --------------------------------------------------------------------- > >> >> To unsubscribe, e-mail: [email protected] > >> >> For additional commands, e-mail: [email protected] > >> >> > >> >> > >> > >> --------------------------------------------------------------------- > >> To unsubscribe, e-mail: [email protected] > >> For additional commands, e-mail: [email protected] > >> > >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
