Re: [Twisted-Python] Interesting plot on new author contributions post-Github for several Python packages
On Tue, Aug 27, 2013 at 2:14 PM, wrote: > On 09:54 am, _...@lvh.io wrote: > >> IIUC, Tom does most of his contributing through Github. That probably >> means >> that it's a well-supported process that has most of the kinks ironed out: >> after all, if Tom hasn't caught them, I would guess a new contributor >> probably wouldn't either :-) >> > > Or it means Tom's workflow is fine and he's learned all the issues he > needs to avoid tripping over. > That sounds like what I mean. I don't understand the difference. I am saying that Tom's workflow is fine and knows where the hard parts are. Also, given the amount of time that he (usually) gets to spend on Twisted and his responsiveness on the IRC channel, I think it counts as "well-supported". > You didn't quote anything in the message you replied to, so I'll just > guess that this is an email meant as a "+1 please switch to github". Please > correct that impression if it is mistaken. > No. I am suggesting that there is a way to contribute using Github that isn't totally broken, and someone has already figured out what that way is, and that someone is in a position to explain it to others. I'm saying we don't need to switch to github, because apparently you can already use github if you want to, everything else staying the same. It's possible that this way of using github does not please the would-be new reviewers sufficiently, but I'm saying that if someone suggests we move to github, I'd like to make it clear to them that you can already use github, and that they should probably at least formulate their arguments in the shape of "the current way that I can use github is bad because ..." :) Thanks, > Jean-Paul cheers lvh ___ Twisted-Python mailing list Twisted-Python@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python
Re: [Twisted-Python] Interesting plot on new author contributions post-Github for several Python packages
On 09:54 am, _...@lvh.io wrote: IIUC, Tom does most of his contributing through Github. That probably means that it's a well-supported process that has most of the kinks ironed out: after all, if Tom hasn't caught them, I would guess a new contributor probably wouldn't either :-) Or it means Tom's workflow is fine and he's learned all the issues he needs to avoid tripping over. You didn't quote anything in the message you replied to, so I'll just guess that this is an email meant as a "+1 please switch to github". Please correct that impression if it is mistaken. If it is not mistaken, then please let this sub-thread of the discussion die right here (reasons for doing so already given in near countless previous threads). The ideas suggested elsewhere for encouraging more reviewers and other contributions seem more likely to bear fruit. Let's follow through on those. Thanks, Jean-Paul ___ Twisted-Python mailing list Twisted-Python@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python
Re: [Twisted-Python] Interesting plot on new author contributions post-Github for several Python packages
IIUC, Tom does most of his contributing through Github. That probably means that it's a well-supported process that has most of the kinks ironed out: after all, if Tom hasn't caught them, I would guess a new contributor probably wouldn't either :-) cheers lvh ___ Twisted-Python mailing list Twisted-Python@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python
Re: [Twisted-Python] Interesting plot on new author contributions post-Github for several Python packages
On 26/08/2013, Terry Jones wrote: > This doesn't prove anything, but I think the first plot at > http://jakevdp.github.io/blog/2012/09/20/why-python-is-the-last/ is quite > provocative. Terry, I don't think the problem is "too few contributors". A big problem is that there are too few reviewers - especially while Tom's away. There are 52 branches currently awaiting review: * https://twistedmatrix.com/trac/report/15 Here are a few thoughts: 1. Encourage new code reviewers -- various people have said they'd do code reviews but are unsure of the process or feel unqualified to comment on other people's code. So: 1. Update the code review checklist * https://twistedmatrix.com/trac/wiki/CodeReview * https://www.ibm.com/developerworks/rational/library/11-proven-practices-for-peer-review/ (via @zooko) * http://mumak.net/stuff/your-code-sucks.html 2. Make the code review checklist and the highscores pages more prominent. Maybe via a new "Contribute" link on the top right of the homepage. (I think ashfall was working on a new wiki page for contributors). 3. Less emphasis in report 15 on "By Order You Should Review Them In" * It's very difficult (for me at least) to review branches that I know nothing about. * Reviewers should be encouraged to review tickets for components that they are familiar with or tickets which they are interested in learning more about. * https://twistedmatrix.com/trac/report/15 2. Encourage code review trading 1. Contributors could be pointed to report 15 after submitting their patch or branch. 2. Encourage contributors to make contact with another developer and agree to trade reviews. 3. Modify report15 so that branches from contributors with highscores get moved higher up the list (http://twistedmatrix.com/highscores). 3. Encourage self review -- It's frustrating when you come to review a patch and you have to point out obvious things. (merge conflicts, no tests, spelling mistakes, missing docstrings, missing documentation) 1. Ask contributors to do a self review before submitting their patch or branch for peer review. 2. Make it easier for contributors to run twistedchecker, pydoctor, pyflakes, lore locally before submitting code for review. 3. Since buildbot is checking out code from Github now, perhaps there could be an interface for contributors to request it to build their own Github branch? 4. Automatically detect merge conflicts in branches and patches awaiting review and notify the author. * This would prevent the frustrating situation of having waited for a month for a review only to have it pushed back to the bottom of the queue with a request to merge forward. 4. Publicise upcoming features / High priority tickets and branches 1. For example, Twisted Web Sockets which people keep asking about on IRC. 2. It's not obvious that there is a Web Sockets server branch which is almost ready to be merged. https://twistedmatrix.com/trac/ticket/4173 3. It's been reviewed over ten times and is now back with therve who may or may not have time to address the latest code review comments. 4. Maybe it would be nice to highlight tickets like this on the twistedmatrix.com front page. That might encourage new people to help push them through to completion. > Semi-related: I made a tiny optimization to jQuery a couple of months ago > as a completely unknown first-time contributor. The time from doing the git > clone to having the pull request merged into jQuery master was about an > hour, which included signing their contributor agreement. You mean this one? * https://github.com/jquery/jquery/pull/1321 The turn-around time can be just as quick in Twisted. Here are a few recent small changes that were merged within an hour or two: * https://twistedmatrix.com/trac/ticket/6674 * https://twistedmatrix.com/trac/ticket/6525 * https://twistedmatrix.com/trac/ticket/6403 One final thing. I hate to hear people beating up Trac. That article follows the trend, stating that Github "replaced the clunky Trac system of submitting static patches to projects". Well that's a gross simplification. I for one am very fond of Trac. I find its UI intuitive -- much more intuitive than Github, and from what I've seen of the code it's quit well designed too. It's a flagship Python application and we should be supporting it. It's just a shame it wasn't built on top of Twisted. Sorry for the long email. -RichardW. ___ Twisted-Python mailing list Twisted-Python@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python
[Twisted-Python] Interesting plot on new author contributions post-Github for several Python packages
This doesn't prove anything, but I think the first plot at http://jakevdp.github.io/blog/2012/09/20/why-python-is-the-last/ is quite provocative. Semi-related: I made a tiny optimization to jQuery a couple of months ago as a completely unknown first-time contributor. The time from doing the git clone to having the pull request merged into jQuery master was about an hour, which included signing their contributor agreement. Terry ___ Twisted-Python mailing list Twisted-Python@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-python