On 03/05/2018 01:29 PM, John D. Ament wrote: > On Mon, Mar 5, 2018 at 1:12 PM Mike Beckerle <mbecke...@tresys.com> wrote: > >> John, >> >> >> This issue is one of the last remaining things we need to resolve. >> >>> >>> 4) Snippets from the OGF DFDL spec. These snippets are scattered >>> throughout multiple files, but are all located in an "ibm-tests" and >>> "ibm-contributed" directories. A grep for "Open Grid Forum" will show >>> find the copyright that is included in all these files. The directories >>> containing the files are in: >>> >>> daffodil-test/src/test/resources/org/apache/daffodil/ibm-tests/ >>> daffodil-test-ibm1/src/test/resources/test-suite/ibm-contributed/ >>> >> >> Same, please list these. Also note that partial files get a bit muddier, >> if there's a way you can keep the files separate that would be make it >> easier (hence why I've asked about modifying the source). >> We can list these files, but decomposing them would lose the whole point >> of them, which is that Daffodil can run them, as is, as they were created >> by IBM, without changes being required. This is an important part of our >> interoperability claims with IBM's DFDL implementation. It's also not >> possible to split up a DFDL schema into multiple files without introducing >> the DFDL language issues associated with multi-file schema composition - >> e.g., namespaces, introduction of new lexical scopes around the contents of >> additional files, etc. Those are important things to test, but are not what >> these particular tests are about, so introducing them would reduce the >> isolation of the tested behaviors. >> >> To me this use of example snippets drawn from the OGF spec document is >> legitimate fair use, and is exactly what was intended by myself and the >> other authors when these examples were included in the DFDL specification. >> >> So how do we satisfy concerns around this matter? Can we simply identify >> exactly the files that contain this sort of material, and explain that >> these files are maintained as-is, in the form originally contributed, so as >> to serve to demonstrate interoperability with other implementations of the >> DFDL standard? >> >> > I'm not sure I understand what you're asking for. My ask is that we list > out the license for the files coming from outside sources. > > I have an additional ask if we have modified the source code. > > Please note that if this is something derived from a specification, then > you're not actually bringing in source code. We care about source code. >
Some of our files are majority written by us (Apache v2) but include snippets that came out of the Open Grid Forum DFDL specification (OGF licensed). We did not make any changes to the snippets. Note that these snippets are XML schema, which could be considered source code, but if not maybe this is all moot? For example, this is one of those files: https://github.com/apache/incubator-daffodil/blob/master/daffodil-test/src/test/resources/org/apache/daffodil/ibm-tests/dpaspc7132.dfdl.xsd In that file, lines 36-43 are an OGF snippet, but the rest of the file is custom written. The header of the schema is Apache v2 license, and line 25 contains a blurb about some content being OGF licensed. PR-51 adds content to the OGF section of the LICENSE file specify these files that contain OGF content. You mentioned that files that contain mixed licenses can get muddy. Mike looked into separating out the OGF snippets and his conclusion is that separation would either be too difficult or would change the intention of a test, so would not be ideal. I guess we would just like clarity if we are properly handling these mixed files with the changes to LICENSE in PR-51 by listing those files in the OGF license section and including the OGF copyright blurb in the files themselves. Thanks, - Steve
