Messages in this thread | | | From | Arnd Bergmann <> | Subject | Re: [Ksummit-discuss] checkkpatch (in)sanity ? | Date | Mon, 29 Aug 2016 11:01:08 +0200 |
| |
On Monday, August 29, 2016 9:15:15 AM CEST Alexandre Belloni wrote: > > > > commiter commits issues > > arnd 858 2155 > > axboe 53 22 > > corbet 15 9 > > davem 55 81 > > grant.likely 2 0 > > gregkh 38 46 > > hch 393 581 > > James.Bottomley 15 15 > > martin.petersen 18 20 > > mchehab 678 1042 > > mgorman 104 256 > > mingo 58 192 > > paulmck 176 68 > > peterz 226 511 > > rostedt 123 178 > > shuahkh 53 6 > > tglx 200 287 > > torvalds 64 89 > > tytso 37 77 > > viro 350 256 > > > > And for the last 10,000 commits in the log, that script has observed 10,783 issues. > > > > It'll be interesting to hear from these people about their view of checkpatch, but IMO when on average there are more issues than commits I can suggest two possible causes: > > > > 1. People are used to ignore checkpatch warnings. > > 2. People aren't using checkpatch. > > > > Well, Arnd is used to move around old code when refactoring. As the code > just moves, he rarely solves checkpatch issues when doing so which is > the right thing to do.
I don't find checkpatch.pl overly useful for my own patches and rarely run it. I looked over the last few hundred commits and found that almost all the warnings were for:
- having overly long lines in commit messages when I quoted a long compiler warning. I generally don't wrap those to make it easier to search for the warnings in the git history
- missing the word "commit" before a reference to another changeset in full-text. I'll change that in the future if that makes people happy, but it doesn't seem important.
- existing style issues that I did not fix when fixing a bug. In many cases I find it better not to change coding style while fixing a bug, but there are other cases in which I do.
Arnd
| |