Re: [PATCH] cleanup: Tweak and re-run return_directly.cocci

[Markus Armbruster]

Peter Maydell <peter.maydell@linaro.org> writes:

> On Tue, 22 Nov 2022 at 08:58, Markus Armbruster <armbru@redhat.com> wrote:
>> I don't think complete detailed review is necessary or even sensible.
>>
>> Review should start with the Coccinelle script:
>>
>>     // replace 'R = X; return R;' with 'return X;'
>>     @@
>>     identifier VAR;
>>     expression E;
>>     type T;
>>     identifier F;
>>     @@
>>      T F(...)
>>      {
>>          ...
>>     -    T VAR;
>>          ... when != VAR
>>
>>     -    VAR = (E);
>>     -    return VAR;
>>     +    return E;
>>          ... when != VAR
>>      }
>>
>> What could go wrong?  Not a rhetorical question!
>
> The obvious answer is "you might have got your manual tweaking
> wrong". A purely mechanised patch I can review by looking at
> the script and maybe eyeballing a few instances of the change;
> a change that is 99% mechanised and 1% hand-written I need to
> run through to find the hand-written parts.

Define "handwritten" :)

If reverting unwanted line-breaks and blank lines counts, then I can
make two patches, one straight from Coccinelle, and one that reverts the
unwanted crap.  The first one will be larger and more annoying to review
than this one.  A clear loss in my book, but I'm the patch submitter,
not a patch reviewer, so my book doesn't matter.

Else, we're down to one file, which I already offered to split off.

> But mostly this patch is hard to review for its sheer size,
> mechanical changes or not. A 3000 line patchmail is so big that
> the UI on my mail client gets pretty unwieldy.

With the manual one split off, target/xtensa/ dropped as requested by
Max, and tests/tcg/mips/ dropped because its status is unclear (and I
start to find it hard to care), we're down to

 28 files changed, 81 insertions(+), 221 deletions(-)

This will be v2.