Re: [PATCH 0/2] checkpatch: Do appropriate kernel->qemu renaming

[Aleksandar Markovic <aleksandar.qemu.devel@gmail.com>]

суб, 20. јун 2020. у 17:09 Aleksandar Markovic
<aleksandar.qemu.devel@gmail.com> је написао/ла:
>
> суб, 20. јун 2020. у 16:25 Peter Maydell <peter.maydell@linaro.org> је
> написао/ла:
> >
> > On Sat, 20 Jun 2020 at 14:33, Aleksandar Markovic
> > <aleksandar.qemu.devel@gmail.com> wrote:
> > > There are several places where 'kernel' is mentioned instead of
> > > 'qemu' in checkpatch.pl.
> > >
> > > This small series corrects this.
> >
> > So, the thing about this kind of change is that it's implicitly saying "we have
> > forked checkpatch and will not try to update to newer versions of it from
> > the kernel tree", because this sort of no-behavioural-change tends to
> > get in the way of that kind of update by obscuring the delta between
> > the kernel version and ours. Effectively I think we *have* ended up
> > with our own fork,
> > simply because we haven't cared to do that kind of update from the kernel's
> > version and we've gradually added our own changes to our copy. But it seems
> > like it's worth discussing the general principle.
> >
>
> I do not say (explicitly or implicitly) that we should or should not
> follow and attempt to mirror changes in checkpatch.pl from kernel. (In
> fact, I think we should.)
>
> However, I don't think that several differences (in this series, I
> thinks altogether 7 lines) that would originate from difference of
> names QEMU vs. kernel would not be any significant obstacle for a
> potential future attempts to do comparison QEMU checkpatch vs kernel
> chekpatch.
>

Sorry, I meant to say:

However, I don't think that several differences (in this series, I
think, altogether 7 lines) that would originate from difference of
names (QEMU vs. kernel) would be any significant obstacle for a
potential future attempts to do comparison QEMU checkpatch vs kernel
checkpatch.



> Take a look at two versions of top_of_kernel_tree below - they already
> differ in their body. Left this way, QEMU's checkpatch version simply
> violates basic naming principles in software development. And, it
> looks we want to sacrifice the principle - just for the sake of
> convenience of a potential engineer having 7 less line in his diff
> (out of much more).
>
> I don't insist on these two patches. I, of course, leave the decision
> to Peter, Paolo, Stefan, Michael, or others tracking kernel's
> checkpatch script.
>
> Thanks,
> Aleksandar
>
> QEMU version:
>
> sub top_of_kernel_tree {
> my ($root) = @_;
>
> my @tree_check = (
> "COPYING", "MAINTAINERS", "Makefile",
> "README.rst", "docs", "VERSION",
> "linux-user", "softmmu"
> );
>
> foreach my $check (@tree_check) {
> if (! -e $root . '/' . $check) {
> return 0;
> }
> }
> return 1;
> }
>
> Kernel version:
>
> sub top_of_kernel_tree {
> my ($root) = @_;
>
> my @tree_check = (
> "COPYING", "CREDITS", "Kbuild", "MAINTAINERS", "Makefile",
> "README", "Documentation", "arch", "include", "drivers",
> "fs", "init", "ipc", "kernel", "lib", "scripts",
> );
>
> foreach my $check (@tree_check) {
> if (! -e $root . '/' . $check) {
> return 0;
> }
> }
> return 1;
> }
>
> > thanks
> > -- PMM