[U-Boot] [PATCH v4] Add 'patman' patch generation, checking and submission script

[Simon Glass]

Hi Albert,

On Fri, Feb 3, 2012 at 12:00 PM, Albert ARIBAUD
<albert.u.boot at aribaud.net> wrote:
> Hi Simon,
>
> Le 03/02/2012 20:30, Simon Glass a écrit :
>
>> Hi Albert,
>>
>> On Fri, Feb 3, 2012 at 11:00 AM, Albert ARIBAUD
>> <albert.u.boot at aribaud.net>  wrote:
>>>
>>> Hi Simon,
>>>
>>> Le 15/01/2012 02:20, Simon Glass a écrit :
>>>
>>>> Hi,
>>>>
>>>> Hmmm patman found a tag in this commit and tried to send it to Fred
>>>> Bloggs. I have added the line below - sorry for the confusion.
>>>>
>>>> On Sat, Jan 14, 2012 at 5:12 PM, Simon Glass<sjg at chromium.org>    wrote:
>>>>>
>>>>>
>>>>> What is this?
>>>>> =============
>>>>>
>>>>> This tool is a Python script which:
>>>>> - Creates patch directly from your branch
>>>>> - Cleans them up by removing unwanted tags
>>>>> - Inserts a cover letter with change lists
>>>>> - Runs the patches through checkpatch.pl and its own checks
>>>>> - Optionally emails them out to selected people
>>>
>>>
>>> It seems to me that this is not specific to u-boot, and as such, has no
>>> reason to be managed along with U-Boot.
>>>
>>> Why should it not simply have its own project and resources?
>>
>>
>> It was created in response to Wolfgang's comments that the sometimes
>> low quality of patches on the U-Boot ML chews up a lot of time.
>>
>> It is not technically specific to U-Boot (since it could also be used
>> with the Linux kernel), but that was my intent when creating it. If
>> this goes in then I plan to look at how to automate build testing also
>> so we can more easily see that patches are MAKEALL-clean.
>>
>> The discussion at the time was here:
>>
>> http://lists.denx.de/pipermail/u-boot/2011-October/105862.html
>>
>> and the many messages which followed. My patch was in response to this.
>
>
> This aptly explains why the tool was created and what it intends to attain,
> and that its use in U-Boot was discussed -- things that I did not question
> in the first place, and still do not.
>
> What I question is the relevance of tying together the development, source
> code management, release management etc of U-Boot and patman. Is there any
> reason that a release of U-Boot should also be a release of patman, or the
> other way around? Is there any reason even to have similar development
> models for both projects?

I'm not sure / don't mind. At little grand to describe this set of
python scripts as a project.

>
> As an example, the creation of git was strongly motivated by, and tied to,
> source code management requirements of Linux, but git is not integrated in
> the Linux source tree and its development is fully independent from that of
> Linux.

Yes, I understand your point, just not sure that I have a strong opinion.

>
> My point is that patman, not being tied to a given project and being
> certainly just as useful to may other git-based projects, it should have its
> own project, be able to evolve at its own pace, etc.

It's use of checkpatch.pl probably reduces the field of play a fair
bit. It was designed for use with U-Boot, but could be used with the
kernel. Perhaps there are others, I'm not sure.

One advantage of having it in the U-Boot source tree tools directory
is that people might find it and use it (at least at the early
stages). If it outgrows its space there due to extra features or more
generic use then it could move perhaps.

Regards,
Simon

>
> Amicalement,
> --
> Albert.