More people with push privileges?

[Sameer Deshmukh]

Hello,

I currently have like 6 Pull Requests pending for NMatrix, so I wondering if more people who have an understanding of the code as much as John can take up reviewing code and merging Pull Requests?

I think more speed in the review + merge process would keep things up to date and also encourage more contributors in this endeavour.

Regards,

Sameer Deshmukh.

[John Woods]

I'm not sure anyone knows the code well enough (even me). I'd be willing to give, say, Colin Fuller push privileges, but I can't think of many other people. Generally, we should have two people looking over every pull request independently anyway.

I do apologize for the delay these last few weeks. It's been the holidays here, and that's disrupted my normal work schedule. I'll try to start looking at PRs tomorrow.

John

--
You received this message because you are subscribed to the Google Groups "SciRuby Development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sciruby-dev...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Colin Fuller]

(sorry for the delay, just finally getting caught up on the post-holidays e-mail deluge)

Push privileges or no, I can help review for sure.  My normal workflow doesn't involve spending a lot of time in the github web UI, so I'm not often totally on top of new pull requests and whatnot.  Feel free to @-tag me on github for a review if you feel like you're not getting a timely response, and that will send me a more aggressive message :)

Cheers,

Colin

[Pjotr Prins]

Even though the main project is slow at picking up pushes, I don't
think that is a concern. Git is meant to be a distributed development
system, so any repo can be a test/staging/whatever version. These
repos should be used for testing and when considered 'great' they can
be merged into the main one.

We are a small community, but it does not mean we need to opt for
centralized code management.

Sameer, you can set up your own repo and become a qualified maintainer
for SciRuby. Just announce your work on the ML. It is a model I would
champion. At some point John may merge your pull requests without
thinking.

It would be good, nevertheless, to allow a second person the power to
merge pull request on the main repo.

Pj.

[Magdalen Berns]

Even though the main project is slow at picking up pushes, I don't
think that is a concern. Git is meant to be a distributed development
system, so any repo can be a test/staging/whatever version. These
repos should be used for testing and when considered 'great' they can
be merged into the main one.

That only works if pull requests are master compatible by the time they get reviewed. If something else newer conflicts then the patch is not going to merge. So, I think Sameer makes a fair point that contributors could feel discouraged. Personally, I have had open pull requests not get reviewed and the same code I wrote get re-authored down the road by someone else and then pushed, which has put me off contributing to the nmatrix gem altogether. If you are keen to stick to the current approach, I would suggest seeing open pull request reviews as a priority before pushing any other new code and if for whatever reason a pull request is no longer master compatible by the time it gets reviewed, it seems appropriate to communicate that to the pull request contributor, in the first instance.

Magdalen

[John Woods]

Yes, Magdalen raises a good point. We currently have this annoying problem where there are like six different forks of Ruby/GSL and no one really knows whose is authoritative. I think that's unlikely to happen with NMatrix, but would rather have a clearly authoritative version — especially with generous submitters like Sameer who are doing an excellent job but still getting a lot of feedback and learning the ropes. (Thanks, Sameer, for putting up with all of my nitpicking.)

John

--

[Sameer Deshmukh]

John if you're OK with Colin being granted push priviledges, could you please do so (provided that hes OK with it too)?

It would really streamline a lot of processes and prevent PRs from becoming so old that they need manual merging, which IMHO would reduce your work overhead too.

[John Woods]

Just did it. Thanks for the reminder.