On Tue, 25 Aug 2020 at 06:57, Thomas Huth <th...@redhat.com> wrote: > > On 24/08/2020 20.46, Peter Maydell wrote: > > We don't mandate SPDX[*], > > but it will do the job if you want to use it. > > > > [*] Mostly because nobody's cared enough to say "we should > > standardize on this and convert existing files and add the > > check to checkpatch that new files have an SPDX line". > > We should standardize on this and convert existing files and add the > check to checkpatch that new files have an SPDX line! :-) > > Ok, now sombody said it loud. Would there be any objections to enforce > this via checkpatch on new files?
I wouldn't object, indeed I think checkpatch-enforcement is an important part of the process. I think that if we're going to move to SPDX we should: * look at the SPDX spec and identify what we need to do beyond just adding SPDX lines (eg do we need a LICENSES/ subdir like the kernel has with a file per license?) * document (in docs/devel?) that we require SPDX * decide what our plan is for 3rd-party code (libvixl, etc) [where any SPDX line we add will be lost again next time we resync our copy of the code] * put in a checkpatch check for new files (presumably the Linux kernel has one we can borrow) * update existing files (I think the kernel folks probably have scripted stuff for the easy parts of this; multi-license files like fpu/softfloat.c likely need by-hand conversion) Mostly I think if we're going to do this we should find somebody who wants to put in the work to push it forwards so we don't have a half-and-half interminably extended transition between old-style license notices and SPDX. thanks -- PMM
