On 31.05.23 11:11, Krzysztof Kozlowski wrote: > On 31/05/2023 09:22, Frieder Schrempf wrote: >> On 31.05.23 08:56, Krzysztof Kozlowski wrote: >>> On 31/05/2023 08:57, Joy Zou wrote: >>>> Update pca9450 bindings. >>>> >>>> Signed-off-by: Joy Zou <joy.zou@xxxxxxx> >>>> --- >>> >>> Subject prefix is: regulator: dt-bindings: pca9450: >> >> Is there some way to have this consistent for all subsystems? Most >> subsystems seem to use: >> >> dt-bindings: [subsystem]: >> >> But some use: >> >> [subsystem]: dt-bindings: >> >> Casual contributors (like me) will very often get it wrong on the first >> try. Examining the history is extra effort that could be avoided and >> often doesn't provide a definite hint as you find both variations in the >> past. >> >> Can we standardize this and make checkpatch validate the subject line? > > I understand your pain. :) > > My expectation is just to have "dt-bindings:" prefix. It can be anywhere > - first or second, doesn't matter to me. > > Then there is the generic rule that subsystem prefix should be the first > and here there is a disagreement between some folks. Most maintainers > either don't care or assume bindings are separate subsystem. Mark (spi, > ASoC, regulator) and media-folks say it is not separate subsystem (real > subsystem are spi, regulator etc), thus they want their subsystem name > as the first prefix. It sounds reasonable. Anyway it does not contradict > DT bindings maintainers expectation to have somewhere "dt-bindings:" prefix. Ok, thanks for the explanation. Would be nice if maintainers could agree on one version then. > > My comment was only to help you and there is no need to resend. I think > Mark when applying will drop "dt-bindings" prefix if is before > regulator, though. Life, no big deal. Im not the patch author, I was just jumping in as I saw your reply and it already happened a few times to me that I needed more than one try and used precious maintainer time just to get the subject right. So I thought there is some potential for improvement. > > Whether checkpatch can do this? Sure, quite likely, one just need some > Perl-foo to add such rule. :) Ok, this isn't something for me, but maybe someone around can come up with an approach.
