https://bugs.linuxfoundation.org/show_bug.cgi?id=1302
--- Comment #3 from J Lovejoy <[email protected]> 2015-07-15 19:46:45 UTC --- (In reply to comment #2) > Unfortunately, the guidelines and data provided make it programmatically hard > to implement reliable matching of candidate license text with the provided > templates. I did find the guideline reference, but this issue was kind of to > test the waters as to how amenable the project was to improving this state of > affairs. > > Things like the Apache license case are very easy for humans, but very > difficult for computers; given that the SPDX project maintains a curated list > of licenses, the solution that makes the most sense, and makes this project > the > most useful, is to maintain the template data in the ideal format for usage in > matching/validation. > > This ideal format would be, essentially, a format that makes the matching > rules > easiest to apply; it could go as deep as to include correct metadata in all > relevant templates or as shallow as to simply remove headers and trailers that > are irrelevant to the substantive license content. > > (There are other problems with the matching rules conflicting with the data, > such as how the matching rules say to ignore Copyright declarations, but > template files such as BSD-3-Clause actually include metadata for the > copyright > declaration, making it impossible to match the template as specified against > data processed according to the guidelines) > > In summary, you might read this report as a request to begin cleaning up the > data in the repository for better processing, beginning with a very simple > case. Is this likely to happen? Glad you found the matching guidelines and have reviewed them thoroughly, some people seem to miss that important piece of information, so I had to at least ask! It is well understood and was a topic of much direct and indirect discussion that the guidelines, in some cases, are easy for a human to understand, but challenging to implement mechanically. We needed to have guidelines that were human-readable and then had a lot of discussion as to how much markup would be used, trying to balance giving tool makers more specific guidance via the markup, but not so much such to constrain implementations. There were some who argued for markup on every license. There was also a desire to avoid markup on every license, as this represented a lot of work and we are generally under-staffed; more markup could always be added later, if it was deemed helpful or necessary. So, yes, we are certainly amenable to suggestions for changes or proposals. It would be preferable to take a more global approach, rather than by-license, though. Of course, if you have found things that you think are contradictory, then we definitely want to know about that too! In any case, the matching guidelines and templates/markup for the licenses are maintained by the SPDX legal team, so any suggestions or proposals need to be brought there (possibly in addition to the tech team). Our next call is Thursday, August 6th - I'm happy to add this to the agenda, if you can join then. thanks, Jilayne -- Configure bugmail: https://bugs.linuxfoundation.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. _______________________________________________ Spdx-tech mailing list [email protected] https://lists.spdx.org/mailman/listinfo/spdx-tech
