[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180510141451.31cb030e@vento.lan>
Date: Thu, 10 May 2018 14:14:51 -0300
From: Mauro Carvalho Chehab <mchehab+samsung@...nel.org>
To: Jonathan Corbet <corbet@....net>
Cc: Christoph Hellwig <hch@...radead.org>,
Linux Doc Mailing List <linux-doc@...r.kernel.org>,
Mauro Carvalho Chehab <mchehab@...radead.org>,
linux-kernel@...r.kernel.org, Ingo Molnar <mingo@...hat.com>,
Peter Zijlstra <peterz@...radead.org>
Subject: Re: [PATCH 13/18] wait: wait.h: Get rid of a kernel-doc/Sphinx
warnings
Em Thu, 10 May 2018 13:42:58 -0300
Mauro Carvalho Chehab <mchehab+samsung@...nel.org> escreveu:
> Em Thu, 10 May 2018 09:38:46 -0600
> Jonathan Corbet <corbet@....net> escreveu:
>
> > On Thu, 10 May 2018 11:21:13 -0300
> > Mauro Carvalho Chehab <mchehab+samsung@...nel.org> wrote:
> >
> > > The problem with a hint-based mechanism is that it will generate
> > > false hints. If added, we may end by needing to add extra tags to
> > > disable the hints mechanism where it gets wrong, or to periodically
> > > do code changes at kernel-doc comments in order to make the hints
> > > logic happy.
> > >
> > > So, IMO, we should provide non-hints based mechanism, like forcing the
> > > string that prepends the colon to have a keyword that will make it to
> > > parse the block as literal, where expressions like:
> > >
> > > See the code-block foo:
> > > See the following code example:
> > > See the following flow diagram:
> > > See the following artwork:
> > >
> > > Is the best alternative to avoid "::", as on the enclosed patch.
> >
> > But this, too, is a hint-based mechanism. Thanks for the patches, I'll
> > keep them around, but I would like an opportunity to try to do better
> > before applying them. I fear that using magic words in this way will
> > lead to a constant stream of surprises, and I'd like to avoid that if
> > possible...
>
> Yes, it is still hint-based. A careful selection of the "magic spell
> words/phrases" would minimize the risks of false positives, but it
> could still lead into some unwanted surprises.
Btw, running this:
$ git grep -A2 "\*\s.*following.*(code|example|artwork|flow|diagram).*:$"
currently doesn't have a single match.
If we force a two word combination, and an ending with ":" should
be enough to not having too much false positives.
Regards,
Mauro
Powered by blists - more mailing lists