Re: [PATCH v4 00/29] Create a book for Kernel development

From: Mauro Carvalho Chehab
Date: Wed Sep 21 2016 - 05:20:45 EST

Em Tue, 20 Sep 2016 18:44:54 -0600
Jonathan Corbet <corbet@xxxxxxx> escreveu:

> On Mon, 19 Sep 2016 08:07:34 -0300
> Mauro Carvalho Chehab <mchehab@xxxxxxxxxxxxxxxx> wrote:
> > That's the 4th version of this series. It also contains a second patch series
> > with more ReST conversions and documentation improvements.
> > This patchset merges the content of a second patch series:
> >
> > [PATCH 00/17] Improve documentation for the development-process
> >
> OK, I'm applying these through 28; I'm going to hold off on #29. Thanks
> for separating that part out so nicely.
> > I opted to keep the patch changing the kernel-docs.txt changes
> > (patch 21/29). The patch is already written and another patch
> > (patch 22/29) depends on it, because there are references to
> > this file at Documentation/HOWTO.
> >
> > It shouldn't be hard to get rid of it, but I'm not sure if worths
> > the effort. As I commented, people might find useful to update
> > it to point to more modern documents. If people won't do it,
> > it can still be removed from the Kernel a the next Kernel version.
> I'll take them for now, since there seems to be interest in doing something
> with this document. I kept the applying-patches one as well. But I do
> think that we need to start being a bit more willing to get rid of musty
> old docs. We don't carry unused code because "it might be useful to
> somebody"; I think we should take the same approach to docs. Out-of-date
> or irrelevant docs are a maintenance burden, and they impose a heavy burden
> on the people the docs are most meant to help...
> A few notes:
> #1 didn't apply, I had to do it by hand. I suspect my late application of
> Marcus's work got in the way there.
> #2 had this:
> > Content-Type: text/plain; charset=true
> ...which threw git am for a loop; I had to fix it manually. What gives
> there?

That's really weird! I did only the usual stuff here... patches created
with git format-patch and C/C added via The
resulting patch is sent via:
git send-email patches/tmp

I double-checked: the patches created are without any Content-Type:.
It is git send-email (git-2.7.4-2.fc24.x86_64) that added those:

Content-Type: text/plain; charset=true
Content-Transfer-Encoding: 8bit

I'll try to investigate what's wrong there or otherwise check if
the upstream version from git's repository works better.

> #4 didn't apply and had to be done by hand.
> #10 (CodingStyle) has a lot of ".. code-block:: c" constructs. Why are
> those needed? We're still using C by default for literal blocks, right?

I opted to keep the code-block there, because, at least on one of the
blocks, we had to use:

.. code-block:: none

Because pygments were doing weird highlights on it. So, for coherency,
I ended by keeping it all along.

> #15 (SecurityBugs) leaves the section numbers in place; did you intend
> that?

Yes. Since we remove the :numbered:, and this document had already
the sections numbered manually, I opted to preserve.

Yet, I don't see anything special there that would justify
numbering. So, I guess we can just remove it.

> #21 (kernel-docs.txt) had the charset=true weirdness
> #28 actually, I balked at applying this one, since it assumes that
> the great renaming is taking place, and that hasn't happened yet.

Oh! Ok, I'll fix this one, removing the rename stuff from the
conversion and resend.

> So actually I only went through #27, but that took a long time - seemingly
> longer than it takes you to create them! :)

Sorry for that. I'll try to make it easier for you next time.

> A few of the patches still have the bare "::" lines in them; I think I'll
> just add a patch to fix those up real quick.