Re: [edk2-devel] License Check - was OvmfPkg: Adding support for bhyve as OvmfPkg/Bhyve

["Leif Lindholm" <leif@nuviainc.com>]

On Thu, Jul 02, 2020 at 13:49:45 +0000, Gao, Liming wrote:
> Leif:
> 
> > -----Original Message-----
> > From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of Leif Lindholm
> > Sent: Thursday, July 2, 2020 6:54 PM
> > To: Laszlo Ersek <lersek@redhat.com>
> > Cc: Rebecca Cran <rebecca@bsdio.com>; edk2-devel-groups-io <devel@edk2.groups.io>; Ard Biesheuvel <ard.biesheuvel@arm.com>;
> > Andrew Fish <afish@apple.com>; Justen, Jordan L <jordan.l.justen@intel.com>; Kinney, Michael D <michael.d.kinney@intel.com>
> > Subject: Re: [edk2-devel] License Check - was OvmfPkg: Adding support for bhyve as OvmfPkg/Bhyve
> > 
> > On Thu, Jul 02, 2020 at 11:27:25 +0200, Laszlo Ersek wrote:
> > > This likely comes from BaseTools commit a4cfb842fca9
> > > ("BaseTools/PatchCheck.py: Add LicenseCheck", 2020-06-12).
> > >
> > > One approach would be to remove "VbeShim.h" from the tracked files under
> > > OvmfPkg, replacing it with a PREBUILD command in the OVMF DSC files.
> > > (Then Bhyve could do the same.)
> > >
> > > However, the generator, namely "VbeShim.sh", is not written in Python,
> > > but in (POSIX) shell, and so it can't be called from PREBUILD (I think
> > > it would break OVMF builds on Windows).
> > >
> > > I don't know what to tell you, other than the blanket license
> > > enforcement from commit a4cfb842fca9 is likely wrong.
> > 
> > *Reads patch*
> > *Figuratively spits coffee all over keyboard*
> > 
> > No, this is not OK.
> > 
> > We *STILL* have no agreed process for accepting non bsd+patent content
> > since we dropped the contribution agreement. I have tried to raise
> > this issue several times in the past, and there has never been any
> > outcome from resulting discussions.
> > 
> > So now I'm going to send out a two-patch set consisting of:
> > - Reverting a4cfb842fca9. (Doing nothing is better than implying that
> >   anything !bsd+patent can currently be added to the tree.)
> > - Deleting the statement in ReadmMe.rst erroneously claiming that the
> >   includion of these other licenses are acceptable until such a point
> >   an active decision has been taken, approved by the community, that
> >   this is permitted.
> > 
> 
> If only bsd+patent is allowed, the checker can be enhanced to check this license only. 
> I don't understand why remove this checker.  

Mainly because that was the easiest thing to do :)

But also because:
- The thread that spawned this also raised the problem of
  machine-generated files.
- I am somewhat unhappy the checker got merged in the first place
  without wider community feedback. BaseTools and its contents are
  used for many repositories (even within TianoCore), and this added
  unconditional check breaks the use for some of those.

Regards,

Leif

> 
> Thanks
> Liming
> > > The generated include file *must* be a ".h" file, otherwise the INF file
> > > reference won't be able to trigger an incremental build, if I understand
> > > correctly. So replacing the ".h" suffix with something else, such as
> > > ".genh" (for "generated header") won't work, I believe.
> > >
> > > Modifying the printf invocations in the generator script to also output
> > > a license header would not be right either, IMO. A license tag makes no
> > > sense (I think) without a copyright (C) statement. And what copyright
> > > (C) notice should we put on a generated file?
> > >
> > > Furthermore, although "ReadMe.rst" in the project root states
> > >
> > > """
> > > Contributions of code put into the public domain can also be accepted.
> > > """
> > >
> > > I don't see how the license check implemented in commit a4cfb842fca9
> > > would accommodate a public domain contribution. (I think it would be
> > > fine to place the the generated header file in the public domain, *if*
> > > (a) we could express that somehow (is there an SPDX tag for that?),
> > > *and* (b) if that would eliminate the need for a (C) notice / authorship
> > > mark.)
> > 
> > Public domain is not an OSI-compatible license:
> > https://opensource.org/node/878
> > 
> > The public domain statement is also one that needs to be re-evaluated
> > in light of the dropped contribution agreement.
> > 
> > /
> >     Leif
> > 
> > > Note that this generator use case is not unique to QemuVideoDxe; see for
> > > example commit 1e9d6b0f98b5 ("OvmfPkg/OvmfXen: Creating an ELF header",
> > > 2019-08-21).
> > >
> > > I've now filed a bug for BaseTools:
> > >
> > > https://bugzilla.tianocore.org/show_bug.cgi?id=2833
> > >
> > > Once that bug is solved -- that is, once we standardize a tag for
> > > marking generated source files as such --, we can update "VbeShim.sh" to
> > > produce the tag, in "VbeShim.h". Then OvmfPkg/Bhyve can do the same.
> > >
> > > Thanks
> > > Laszlo
> > >
> > 
> > 
>