Re: Patch priority in subjects ?

Re: Patch priority in subjects ?

[Joe Perches]

On Sun, 2014-07-20 at 19:13 +0200, Fabian Frederick wrote:
> I was reading all those "friendly" messages around checkpatch -f lately
> and the fact that code clean-up is too noisy (?)
> 
> I guess I'm not the first to think about it but why don't we use something
> like a priority field in patch subject ?
> 
> Of course it would be arbitrary but maybe better than nothing ?
> 
> eg
> 
> [PATCH 1/1 0] Urgent bug fix
> [PATCH 1/1 1] Bug fix
> [PATCH 1/1 2] ...
> [PATCH 1/1 7] kernel-doc fix
> [PATCH 1/1 8] Code clean-up
> [PATCH 1/1 9] Trivial fix
> 
> Maybe this could help some people to sort/filter/delete
> 
> It's just an idea of course...

It's a good idea, but this is more a subject for
discussion on and by the list than off-list so I'm
adding LKML.

As far as I know, the only infrequently used prefixes
today are "RFC" and "trivial".

If you start to use a prefix, please use readable
text and not some numeric table index.

[RFC PATCH]
[URGENT PATCH]
[BUGFIX PATCH]
[TRIVIAL PATCH]
[CODE STYLE PATCH]

etc...

Getting people to use
	git format-patch --subject-prefix="text"
will be the useful part of this challenge.

Maybe you could develop YA little helper script for
that, ideally with versioning support for repeated
patch submissions too.

Re: Patch priority in subjects ?

[Richard Weinberger]

On Sun, Jul 20, 2014 at 7:36 PM, Joe Perches <joe@perches.com> wrote:
> On Sun, 2014-07-20 at 19:13 +0200, Fabian Frederick wrote:
>> I was reading all those "friendly" messages around checkpatch -f lately
>> and the fact that code clean-up is too noisy (?)
>>
>> I guess I'm not the first to think about it but why don't we use something
>> like a priority field in patch subject ?

Code clean up is not bad per se.
Actually cleaning up things is very much welcome.

The thing is that pure white space or search&replace patches
produced by checkpatch.pl are often not helpful at all.
They pollute the git history, introduce merge conflicts, add
maintenance overhead, etc..

As somebody who regularly stares at code to find outhow the heck
some line of code got introduced these patches become a major PITA.
Running commands like are needed far too often:
git blame <sha1-of-stupid-cleanup>~1 foo/bar.c

Btw: Don't even try to tell me about the -w switch...

>> Of course it would be arbitrary but maybe better than nothing ?
>>
>> eg
>>
>> [PATCH 1/1 0] Urgent bug fix
>> [PATCH 1/1 1] Bug fix
>> [PATCH 1/1 2] ...
>> [PATCH 1/1 7] kernel-doc fix
>> [PATCH 1/1 8] Code clean-up
>> [PATCH 1/1 9] Trivial fix
>>
>> Maybe this could help some people to sort/filter/delete

No need to add more bureaucracy, such filtering can perfectly done
by looking at the sender name. (Sadly...)
We also have the trivial tree.

-- 
Thanks,
//richard

Re: Patch priority in subjects ?

[Fabian Frederick]

> On 21 July 2014 at 13:33 Richard Weinberger <richard.weinberger@gmail.com>
> wrote:
>
>
> On Sun, Jul 20, 2014 at 7:36 PM, Joe Perches <joe@perches.com> wrote:
> > On Sun, 2014-07-20 at 19:13 +0200, Fabian Frederick wrote:
> >> I was reading all those "friendly" messages around checkpatch -f lately
> >> and the fact that code clean-up is too noisy (?)
> >>
> >> I guess I'm not the first to think about it but why don't we use something
> >> like a priority field in patch subject ?
>
> Code clean up is not bad per se.
> Actually cleaning up things is very much welcome.
>
> The thing is that pure white space or search&replace patches
> produced by checkpatch.pl are often not helpful at all.
> They pollute the git history, introduce merge conflicts, add
> maintenance overhead, etc..
>
> As somebody who regularly stares at code to find outhow the heck
> some line of code got introduced these patches become a major PITA.
> Running commands like are needed far too often:
> git blame ~1 foo/bar.c
>

I understand your point of view now.
Maybe Joe has an idea about this (get_maintainer ...) ?

> Btw: Don't even try to tell me about the -w switch...
>
> >> Of course it would be arbitrary but maybe better than nothing ?
> >>
> >> eg
> >>
> >> [PATCH 1/1 0] Urgent bug fix
> >> [PATCH 1/1 1] Bug fix
> >> [PATCH 1/1 2] ...
> >> [PATCH 1/1 7] kernel-doc fix
> >> [PATCH 1/1 8] Code clean-up
> >> [PATCH 1/1 9] Trivial fix
> >>
> >> Maybe this could help some people to sort/filter/delete
>
> No need to add more bureaucracy, such filtering can perfectly done
> by looking at the sender name. (Sadly...)
> We also have the trivial tree.

I've made a small RFC script so we'll be able to talk around something concrete.
I know it would add some rules to follow but having such a script related in
SubmittingPatches
could help in the following:

-Patch prefix could help making statistics.
-Having automatically -s flag (+others?) avoid thousand of replies "Please add
Signed-off-by".
-Avoid different formats like [PATCH 1 V3] , [PATCH 1/1 v3], [PATCH V3]
-Batch patch generation, checking, sending.
-...

Regards,
Fabian
>
> --
> Thanks,
> //richard