On domingo, 29 de setembro de 2013 20:17:09, Knoll Lars wrote: > I don't fully agree with this. > > I certainly don't want us to add huge amounts of data into our > repositories (as e.g. high resolution vector images). IMO it simply makes > more sense to clearly state how you can get the data from upstream.
I didn't say "add to the repositories". But we need to have a copy of them in our infra. We could just add the files to our download server, in a special area. > In addition, I prefer not to copy upstream data if we can avoid it. > Duplicating upstream data can also be harmful. I've seen patches appear in > that data before that then do not get submitted back upstream (e.g. in our > copy of libpng or harfbuzz). Agreed. If possible, don't have third-parties in the first place. Require them from the system. Unfortunately, that reasonable request is not reasonable on Windows. > At most we should have a readonly copy of the sources on our FTP server, > but I am not sure this really gives any benefits if what we use is a > straight copy of an upstream package. That's what I am asking. The benefit is that we still have the original sources, should the upstream disappear and close down. Also, I believe this is required in order to fulfill our obligations under the LGPL: that's why every company shipping an Android device must have a copy of the kernel sources in their servers and it's not enough to point to kernel.org. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Development mailing list Development@qt-project.org http://lists.qt-project.org/mailman/listinfo/development
