Re: Bottleneck problem to merge Python patches

[Eduardo Habkost]

On Sat, Sep 19, 2020 at 01:17:06PM +0200, Philippe Mathieu-Daudé wrote:
> Hi Aleksandar,
> 
> (extending the Cc list to broader audience)
> 
> On 9/18/20 10:33 PM, Aleksandar Markovic wrote:
> > 
> > On Friday, August 28, 2020, Ahmed Karaman <ahmedkhaledkaraman@gmail.com
> > <mailto:ahmedkhaledkaraman@gmail.com>> wrote:
> > 
> >     - Apply pylint and flake8 formatting rules to the script.
> >     - Use 'tempfile' instead of '/tmp' for creating temporary files.
> > 
> >     ---
> > 
> > Hello, folks.
> > 
> > This series seems forgotten. Can some of you perhaps take a look,
> > review, and possibly integrate some patches in a pull request?
> > 
> > Ahmed invested quite of time to improve the functionality and quality of
> > scripts, and they are truly useful for developers interested in
> > performance measurement.
> 
> The Python patches are indeed being merged slowly, but are not
> forgotten :) Eduardo sent just a pull request *yesterday* for
> patches he had queued *before* QEMU 5.1, that is more than 1 year
> ago!
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg742228.html
> 
> I hope he will be able to process the other Python patches sent
> the last 12 months. He raised the problem few month ago saying he
> was overwhelmed and was looking for help from the community.
> Cleber helped a pair of times, I helped once, but then nobody
> popped up volunteering.

The patches were from July this year.  But, yeah, I'm not being
able to keep up with the maintainer role for Python code, and
plan to change my M: line to R: in the "Python scripts" section.

> 
> I agree this is a community problem, nobody wants to become the
> de-facto Python maintainer. Current maintainers are already busy
> solving problem with low-level languages such C.
> As a project, QEMU is evolving, using more and more Python, switched
> to Meson, we might have Rust code too. Learning that for current
> maintainers takes time. I guess we lack new contributors/maintainers
> with interest in Python & QEMU.

I believe the solution here is to not have a single owner for all
Python code, but owners for specific areas.  We already have 36
F: entries in MAINTAINERS for files inside `scripts/`, and each
of those maintainers can send pull requests containing Python
code in the areas they maintain.  Including Ahmed, who is already
listed as the maintainer of `scripts/performance/`.

> 
> This is my simple/rough analysis, as John had the same problem
> 2/3 months ago, his patches were on the list unreviewed for various
> weeks. Same problem with Avocado patches, Lukas sent a series a bit
> before Ahmed and it is still unreviewed:
> https://www.mail-archive.com/qemu-devel@nongnu.org/msg737272.html
> 
> Alex Bennée recently said:
> 
>  "review time is the currency of qemu-devel ;-)"
> 
> We might be failing at explaining new contributors the importance
> of peer-review, as this helps a lot maintainers. It is described
> in the wiki but maybe new contributors don't read it, we don't
> remember them to:
> https://wiki.qemu.org/Contribute/SubmitAPatch#Participating_in_Code_Review
> and:
> https://wiki.qemu.org/Contribute/SubmitAPatch#Return_the_favor
> 
> My 2 cents...
> 
> Regards,
> 
> Phil.
> 
> > 
> > Thanks,
> > Aleksandar
> 

-- 
Eduardo