Messages in this thread | | | Subject | Re: [PATCH] checkpatch: Extend CamelCase types and ignore existing CamelCase uses in a patch | From | Joe Perches <> | Date | Wed, 04 Sep 2013 16:08:06 -0700 |
| |
(sending for 3rd time, odd dns problems today, apologies for dupes)
On Wed, 2013-09-04 at 08:58 -0700, Sarah Sharp wrote: > On Tue, Sep 03, 2013 at 10:25:21AM -0700, Joe Perches wrote: > > Extend the CamelCase words found to include structure members. > > > > In https://lkml.org/lkml/2013/9/3/318 Sarah Sharp (mostly) wrote: > > > > "In general, if checkpatch.pl complains about a variable a patch > > introduces that's CamelCase, you should pay attention to it. > > Otherwise, [] ignore it." > > > > So, if checking a patch, scan the original patched file if it's > > available and add any preexisting CamelCase types so reuses do > > not generate CamelCase messages. [] > Thanks! Will this mean checkpatch.pl still complains on CamelCase names > if it's run against a file? I think that's still valuable.
Yes.
First, checkpatch looks for all existing CamelCase #defines, typedefs, function names and struct/union members in the include path. (it uses regexes so it's actually not at all close to even good at finding those).
It stores all those CamelCase uses in a hash.
If checkpatch is scanning a patch, it'll now read the file being patched for existing uses of CamelCase #defines, etc, and checkpatch adds those uses to the hash.
If checkpatch is scanning a file, it doesn't doesn't prescan the file.
Then, when checkpatch scans the patch or file and finds a CamelCase use, it looks for that use in the hash and is silent if it's there, noisy otherwise.
This can still report CamelCase uses in a patch if say a CamelCase type is defined in a .h file in the same directory or some other include path and that word is not already used by the file.
| |