Re: Coding style: mixed-case

[Nick Piggin]

Kenneth Aafløy wrote:
> On Wednesday 06 April 2005 04:09, Matt Mackall wrote:
>
> > While there may be reasons why mixed case is suboptimal, the real
> > reason is that it's hard to keep track of which style is used where.
> > It's annoying and error-prone to have to remember the naming format
> > for everything in addition to its name. As most things are in a
> > standard style, things are made easier by having every piece of new
> > code follow that style and let us slowly approach uniformity.
>
>
> My primary concern was that of; why does the kernels own coding style
> deviate from that advise given in it's documentation. Other than that

Probably it's been like that for a long time, and nobody has
really bothered to change it.

> > If you posted a patch for pf_locked() and friends (and note that it's
> > lowercase to match function-like usage), you'd probably find some
> > enthusiasts and some naysayers. Most of the naysayers would object on
> > the grounds of "it ain't broke", but if someone were to do it as part
> > of a series of more substantial clean-ups, it'd likely be accepted.
>
>
> Certainly I would like to have a go at a patch, but I must say that I do not
> feel particularly familiar with the code in question to make such a change.
> I would have risen to the challenge had this been a driver level change,
> but the mmu is something that I will not touch untill I feel comfortable.

Well the only patch that could possibly be considered would be a
straight search and replace, and absolutely no functional changes;
I think you would be up to it ;)

A few suggestions:
Don't use PF_*. That namespace is already being used by at least
process flags and protocol flags. Maybe page_locked, page_dirty,
etc. might be better

There could be a quite a bit of external code using these interfaces.
Typically we wouldn't just rename public interfaces in a stable
series "just because", but the rules are a bit different for 2.6.

Your best bet would be to firstly do a patch to create the new interface
names but keep the old ones in place for backwards compatibility (just
#defined to the new name), then a second patch to convert over all the
in-kernel users. The compatibility stuff can be removed in N years.

Lastly, it is quite likely that many people will consider this to be
more trouble than it's worth. So keep in mind it is not guaranteed to
get included.

--
SUSE Labs, Novell Inc.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/