On 6/9/10 14:52, Justin Edelson wrote:
Richard- Perhaps this is supposed to be obvious, but I think it would be helpful to define the term "uses" with respect to the DEPENDENCIES file. IIUC, it includes dependencies (in any scope) as well as software executed as part of the build (i.e. Maven Plugins), but the inclusion of the latter may not be intuitive.
True. We need to be clearer... -> richard
Justin On Wed, Jun 9, 2010 at 1:55 PM, Richard S. Hall<[email protected]>wrote:With the latest release of the framework and Gogo modules, we've tried to update the release process for how we handle the NOTICE file. Our past usage is apparently not aligned with the intended usage, where the NOTICE file should only contained notices for included third-party software whose license requires notice. Our new approach (for now) is this: 1. Include a NOTICE file which contains only required notices. 2. Include a DEPENDENCIES file which contains our acknowledgments about the software the subproject uses. For the most part, this isn't a major hassle and largely boils down to this: 1. Rename the current NOTICE file to DEPENDENCIES. 2. Create a new NOTICE file that contains notices only for the "used" software requiring notices in the DEPENDENCIES file. Although the new DEPENDENCIES file is very similar to the old NOTICE file, the template is slightly different as indicated here: https://cwiki.apache.org/confluence/display/FELIX/DEPENDENCIES+file+template+%28PROPOSED%29 Of course, in the long term we can try to move to automating the generation of the NOTICE and/or DEPENDENCIES files, which would make our lives simpler. If any subprojects currently are able to automate this information, as long as the generated files contain information consistent with what is proposed here, then the exact formatting is not that important. But for hand generated files, follow this template. If you want to see examples, look in the framework or gogo subprojects. -> richard p.s. This is obviously all open for discussion to the specifics, but until then we should use this approach for releases in an effort to better align with Apache process (with perhaps the exception of Karaf since if/when it goes TLP then its PMC will decide how to do releases).
