Please stop fixing easy issues right now! Leave them as exercices to newcomes
Hi,
I discussed with Mariatta and Carol at Pycon US about new contributors and the difficulty to find "easy issues" to start contributing to CPython. The thing is that easy issues usually are fixed in less than 24 hours which doesn't give the opportunity to newcomers to fix them.
*Many* people ask me regulary "how to find easy Python issues", and the last 3 years, I always failed to find such issues... Many "easy issues" are older than 3 years old, have more than 20 comments and no compromise has been found how to fix the "easy" issue...
I propose a new policy for core developers: stop fixing really easy issues! I suggest to follow Brett Cannon's example. Instead of fixing importlib bugs, Brett told me that he started to describe the bug and explain how to fix it. According to his own experience, it works well and is very valuable!
The plan is to recruit new contributors and mentor them to grow our team. More contributors = more people to review changes = more diversity = less bugs = etc. (long win-win list ;-))
I started with 4 issues on reference leaks found by new Zachary's "Refleak" Gentoo and Windows buildbots. I used a script that I wrote to identify one test leaking references, but I didn't write the fix. Usually, writing the fix is the simplest task: the boring and complex task is more to isolate the leaking test method. Hum, the next step is to explain how to fix such issue, I may do that on the core menthorship mailing list.
http://bugs.python.org/issue30547 http://bugs.python.org/issue30546 http://bugs.python.org/issue30542 http://bugs.python.org/issue30536
I added [EASY] in the issue title to advertise these issues and used the "easy (C)" keyword. Since the proposal rule asking core dev is new, I also write a comment to explain my plan ;-)
More generally, I now suggest to spend more time on mentoring newcomers and review changes instead of writing new changes. This idea is not mine, it's just a very good advice that Mariatta gave me ;-)
Since I know that it's a very different job and can be seen as less interesting, it's not mandatory at all! It's just an advice if you want to try "something new" ;-)
I will also try to spend more time next weeks on our core menthorship mailing list.
What do you think?
Victor
Hi,
Le 02/06/2017 à 11:23, Victor Stinner a écrit :
*Many* people ask me regulary "how to find easy Python issues", and the last 3 years, I always failed to find such issues... Many "easy issues" are older than 3 years old, have more than 20 comments and no compromise has been found how to fix the "easy" issue...
In that case, it's probably reasonable to remove the "easy" tag ;-)
I propose a new policy for core developers: stop fixing really easy issues!
That's a good policy. I remember doing so some years ago. Of course, if some "easy" issue you care about hasn't been fixed for 6 months, perhaps you can fix it yourself after all.
Also, if some such issues are still unfixed at the eve of a release, better fix them yourself too.
Regards
Antoine.
2017-06-02 11:28 GMT+02:00 Antoine Pitrou <antoine@python.org>:
In that case, it's probably reasonable to remove the "easy" tag ;-)
Right, we need to cleanup this old list to "easy" issues.
That's a good policy. I remember doing so some years ago. Of course, if some "easy" issue you care about hasn't been fixed for 6 months, perhaps you can fix it yourself after all.
Also, if some such issues are still unfixed at the eve of a release, better fix them yourself too.
Sure. But we are closer to the beginning of the 3.7 cycle than the end, so it should be ok. Moreover, I noticed more active contributors since we migrated to GitHub. So I don't worry at all :-)
Victor
On Fri, Jun 2, 2017 at 12:23 PM, Victor Stinner <victor.stinner@gmail.com> wrote:
*Many* people ask me regulary "how to find easy Python issues", and the last 3 years, I always failed to find such issues... Many "easy issues" are older than 3 years old, have more than 20 comments and no compromise has been found how to fix the "easy" issue...
See http://psf.upfronthosting.co.za/roundup/meta/issue605 for the previous discussion on easy issues. I re-triaged some of them (the initial list is at http://psf.upfronthosting.co.za/roundup/meta/msg3169) in the past year, but that's not something I want to do in my free time anymore (and unsurprisingly, companies aren't interested to fund issue triaging work)
I added [EASY] in the issue title to advertise these issues and used the "easy (C)" keyword.
Please consider not using prefixes in issue titles. Setting appropriate fields in the issue detail page should be enough. There is no need to duplicate the information in the title.
--Berker
participants (3)
-
Antoine Pitrou -
Berker Peksağ -
Victor Stinner