On Tue, Sep 17, 2019 at 9:16 AM Jason Gunthorpe <jgg(a)ziepe.ca> wrote:
On Fri, Sep 13, 2019 at 02:48:50PM +0300, Dan Carpenter wrote:
> It used to be that infiniband used "sizeof foo" instead of sizeof(foo)
> but now there is a new maintainer.
These days I run everything through checkpatch and generally don't
want to see much deviation from the 'normal' style, a few minor
clang-format quibbles and other check patch positives excluded.
This means when people touch lines they have to adjust minor things
like the odd 'sizeof foo' to make it conforming.
Like others there is a big historical mismatch and the best I hope for
is that new stuff follow the cannonical style. Trying to guess what
some appropriate mongral style is for each patch is just a waste of my
I also hold drivers/infiniband as an example of why the column
alignment style is harmful. That has not aged well and is the cause of
a lot of ugly things.
> There is one subsystem where the maintainer is super strict rules that
> you can't use "I" or "we" in the commit message. So you
can't say "I
> noticed a bug while reviewing", you have to say "The code has a
Ah, the imperative mood nitpick. This one is very exciting to explain
to non-native speakers. With many regular submitters I'm still at the
"I wish you would use proper grammer and sentence structure" phase..
These days I just end up copy editing most of the commit messages :(
> I don't think it's shaming, I think it's validating. Everyone just
> insists that since it's written in the Book of Rules then it's our fault
> for not reading it. It's like those EULA things where there is more
> text than anyone can physically read in a life time.
Yeah, I tend to agree.
The big special cases with high patch volumes (net being the classic
example) should remain special.
But everyone else is not special, and shouldn't act the same.
The work people like DanC do with static analysis is valuable, and we
should not be insisting that those contributors have to jump through a
thousand special hoops.
I have simply viewed it as the job of the maintainer to run the
process and deal with minor nit picks on the fly.
Maybe that is what we should be documenting?
In theory, yes, in practice, as long as there is an exception to the
rule, it comes down to a question of "is this case special like net or
not?". I'd rather not waste time debating that on a per-subsystem
basis vs just getting it all documented for contributors.
I do think it is worth clarifying in the guidelines of writing a
profile to make an effort to not be special, and that odd looking
rules will be questioned (like libnvdimm statement continuation), but
lets not fight the new standards fight until it becomes apparent where
the outliers lie.