On Wed, Mar 20, 2013 at 03:54:32PM -0700, Joe Perches wrote:
On Wed, 2013-03-20 at 17:36 +0100, Mark Brown wrote:
This is just like any other coding style thing - you should be creating patches that look like other patches for the affected, if there's things like obvious visual differences in what you're doing you're doing it wrong.
We've had this conversation before and I proposed to you a simple solution. https://lkml.org/lkml/2010/11/16/245
If you want a script feel free to write one, as repeatedly discussed (including in that thread) it's not completely trivial. Personally I don't feel it's a useful use of time and it's certainly not something I'd have any intention of using.
and I still more or less agree with Florian https://lkml.org/lkml/2010/11/16/314
What he's saying there is that maintainers should just hand edit the patches; that's just stupid especially for trivial patches where all that should be needed is a git am run. You're doing this a lot, you should be getting it right. First time and occasional submitters tend to get a lot more leeway but when submitters send a lot of patches but continually ignore feedback...
I'm not doing it wrong. You have another demand others don't. I simply don't find it necessary to cater to you.
...or even actively reject it then it shouldn't be a surprise when the strength of the pushback ends up increasing.
If you want it to be agreed that there is a specific form for subject headers that varies by maintainer tree, change SubmittingPatches Paragraph 11.
This is all pretty basic stuff (and if it were going to be spelled out in more detail it'd be along with all the other stuff about writing good subject lines). Like I say it's also most important to frequent submitters and not something it's essential to get right first time.
Automation doesn't work for things like this, there's a good solid reason why there's generally a human involved in patch; the other people who submit lots of cleanups generally manage to figure this out usefully, you might want to discuss techniques with them.
I suggest you use a git pre-commit hook to your tree and use sed/perl to add a specific prefix if it doesn't exist. http://codeinthehole.com/writing/tips-for-using-a-git-pre-commit-hook/
I don't think you're quite understanding the issues with automation here. Or indeed the desired end result.