[go: nahoru, domu]

Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Expert teams #2826

Closed
daler opened this issue Nov 7, 2016 · 52 comments
Closed

Expert teams #2826

daler opened this issue Nov 7, 2016 · 52 comments

Comments

@daler
Copy link
Member
daler commented Nov 7, 2016

I was thinking it would be useful to have easy ways of calling in help for especially tricky recipes. Currently the common method is to ping @bioconda/core, but we have a lot of expertise spread across other contributors who could provide better help than @bioconda/core, but aren't necessarily notified.

I'd like to propose a small set of "teams" that can be notified for specific help. Here's a first pass:

  • @bioconda/perl for general Perl recipes. Folks on this team would have experience building Perl recipes and know about the issues we have with perl vs perl-threaded.
  • @bioconda/lib for things like troubleshooting recipes where zlib can't be found. For example folks on this team would know about the kinds of env vars (CFLAGS, LDFLAGS) are useful get things to work.
  • @bioconda/build-system for build-system issues (this is likely redundant with @bioconda/core, but maybe we could get more people helping out here), this will also include help on travis-ci jobs, like why a job mysteriously failed (e.g. HTTP 500 errors or the out-of-disk-space errors we've been seeing recently). Folks here could also act as triage for directing to other expert teams
  • @bioconda/r for R-related issues.
  • @bioconda/haskell for Haskell-related issues.
  • @bioconda/review to ping for a general review of the package - pinning/skipping/metadata ...
  • @bioconda/osx for OSX-specific issues. Folks on this team should have access to a Mac they can use for troubleshooting.

I can think of 2 or 3 people not on bioconda/core for most of these groups, so I think there's a fair amount of expertise we're missing out on (if those people are not already monitoring all PRs and issues). The only cost would be extra emails in team members' inboxes, but the benefit could be better recipes and quicker issue resolution to reduce contributor frustration.

@johanneskoester
Copy link
Contributor
johanneskoester commented Nov 7, 2016

This is a great idea, Ryan! Let's do that. If the core agrees, we could start here, composing a list of users for each team, and ask them if they want to do that.

@daler
Copy link
Member Author
daler commented Nov 10, 2016

Going by contributions/conversations I can remember and a bit of git-log grepping, here's a first attempt. I'm sure I'm missing others, so please add to the list.

team possible members not already on core team
Perl @jerowe, @druvus, @acaprez
gcc/lib @kyleabeauchamp, @yesimon, @druvus, @acaprez, @marcelm
travis/build @kyleabeauchamp
R @lecorguille, @dyusuf, @jerowe
osx @acaprez, @yesimon

As each team identifies other bioconda members with similar expertise they could grow their team to help spread the work.

Anyone listed here so far interested in being on an expert team? It would mean monitoring your email for notifications when someone needs help with your area of expertise, and helping them solve the issue.

@dyusuf
Copy link
Contributor
dyusuf commented Nov 10, 2016

@daler thanks for the effort! yes, I am in.

@martin-raden
Copy link
Contributor

not sure if I am really considable as a 'perl-expert' since I 'chose' accidentially the update of bioperl to v1.7+ as my bioconda intro... kept me busy for more than a week now since this required an immense dependency update. still not done.. :/

but since I had to work around a few quirks, you can ping me if needed for perl-questions.

btw: avoid perl .. ;-)

@acaprez
Copy link
Contributor
acaprez commented Nov 10, 2016

Likewise, I'm not sure I'd consider myself an expert on any or all of those, but I'm happy to contribute and help.

@daler
Copy link
Member Author
daler commented Nov 10, 2016

Thanks @dyusuf @martin-mann @acaprez!

@jerowe
Copy link
Contributor
jerowe commented Nov 11, 2016

I'm in. Always glad to be a part of bioconda. ;-)

@yesimon
Copy link
Contributor
yesimon commented Nov 16, 2016

I'm not an expert but I'll be willing to help in things like OS X

@lecorguille
Copy link
Member

Ouha ... really old thread.
I miss this mention. I will be glad to be part of the R team if bring my modest contribution (and sure, learn stuff)

@bgruening
Copy link
Member

@lecorguille cool I added you.

@bgruening
Copy link
Member
bgruening commented Jan 14, 2018

@bioconda/all happy new year! Anyone wants to join one of the mentioned teams for 2018 and beyond? :)

@bgruening
Copy link
Member

I added also a new group for haskell and added @eggzilla to it :)

@asntech
Copy link
Contributor
asntech commented Jan 14, 2018

@bgruening I'm happy to join the r and osx teams. :)

@tiagoantao
Copy link
Member

Any further thoughts on the list of topics? One thing that crossed my mind is JavaScript/node for stuff like bionode, but people more deeply involved might have a better idea of what is needed...

@bgruening
Copy link
Member

@asntech thanks I have added you.
@tiagoantao I know that a lot of node stuff was added to conda-forge, so I guess with the first JS packages popping up here, this makes a lot of sense. We probably also need a rust team. I will wait for @johanneskoester who has contributed a few rust packages to comment on this.

@jdblischak
Copy link
Member

Happy new year @bgruening! Could I please be added to the R team?

@bgruening
Copy link
Member

@jdblischak awesome, added you!

@acaprez
Copy link
Contributor
acaprez commented Jan 15, 2018

I'd be happy to join the lib team and help out.

@bgruening
Copy link
Member

@acaprez done - thanks!

@cbrueffer
Copy link
Member

I can help out in lib and r.

@johanneskoester
Copy link
Contributor

@bgruening a rust team would indeed be reasonable. For now, I don't think there is so much out there yet (mainly my own stuff), but this will for sure increase.

@bgruening
Copy link
Member

@cbrueffer add you, thanks a lot!
@johanneskoester created and added you :)

@dpryan79
Copy link
Contributor

@bgruening I'm happy to help with gcc/lib

@dlaehnemann
Copy link
Member

@bgruening I'm not very experienced with it, but working on some Rust stuff. So feel free to also add me in there.

@bgruening
Copy link
Member

@dlaehnemann, @daler done! Thanks a lot!

@johanneskoester
Copy link
Contributor

@bgruening I have added @dlaehnemann to the rust team. Did you accidentally add yourself?

@martin-raden
Copy link
Contributor

@bgruening I am sorry, but I have to withdraw my perl-team membership, since I wont have time within the next months. When things are back to normal I will let you know!
Thanks,
Martin

@epruesse
Copy link
Member

I'll be happy to help with gcc/lib, build and osx.

@epruesse
Copy link
Member

The list of help options should probably be somewhere prominent, such as the PR template and the package-build-howto.

@epruesse
Copy link
Member

Maybe another team @bioconda/review ?

@tiagoantao
Copy link
Member
tiagoantao commented Jan 15, 2018 via email

@bgruening
Copy link
Member

@martin-mann I removed you from the group. Thanks for your help!
@epruesse added you.
@johanneskoester I might help with rust as well if it is low level :)

How should be part of @bioconda/review? I like the idea, but it is not clear when to call this group and how should be part of it? We do not have a strict guideline to wait for reviews afaik.

@dpryan79
Copy link
Contributor

@bgruening I though there used to be the idea that if you were submitting a brand new recipe that the core team should give it a thumbs up before merging. I imagine a /review team could serve that purpose (for new recipes that are already passing tests).

@Slugger70
Copy link
Contributor
Slugger70 commented Jan 16, 2018 via email

@epruesse
Copy link
Member

How should be part of @bioconda/review?

I'd start with those that do the reviews and merges most of the times, plus maybe a few volunteers.

I like the idea, but it is not clear when to call this group and how should be part of it?

It would be the group to ping when a recipe is ready for review & merge. It's easier on the new folk than pinging core, and I presume it's easier on you not to have to constantly monitor the PRs for ones that are building ok.

We do not have a strict guideline to wait for reviews afaik.

We claimed in the paper that we enforce a 4-eyes process, and as Bioconda's membership grows, we should move toward a setup where not everyone can just merge themselves. => The members of that group should at some point become the only ones with merge privileges.

@mbargull
Copy link
Member

We do not have a strict guideline to wait for reviews afaik.

I would be very much in favor to make reviews mandatory for every merge. Even for stupidly simple version bumps -- typos or forgotten build number resets and such can always happen.
If one wants to argue this would needlessly slow down package releases, I'd counter that rarely none of those lovely "blazing-fast-never-sleeping-maniacs" are around 😉.

I think having a @bioconda/review might be a good idea, just to have the ability to somehow ping people for a review. The only downside I see is that, IIRC, there is no possibility to "hop-onto/hop-off of" GitHub teams easily. Meaning, it would be nice to have the possibility to join a team, esp. @bioconda/review, only temporarily when one has a bit time to spare (and doesn't have to bug anyone with appropriate permissions on every join/leave).

@bgruening: I can help with anything Python or Conda (and generally Python packaging related), so @bioconda/build would fit. If you don't want to create such a team due to redundancy, I would also be willing to pledge to behave if I'd be added to @bioconda/core. If created, happy to join @bioconda/review as well.

@chapmanb
Copy link
Member

-1 on mandatory reviews. My experience with this has been that it overwhelms the team doing reviews and leads to slow turn around times. conda-forge processes are currently similar to this; my recent PRs there have gotten 3+ days to review and merge for simple version bumps. Even the current slow Travis build times make it hard to respond to issues in tools I'm maintaining that require updating or fixing packages, so I'd find it hard to have a slower process.

@mbargull
Copy link
Member

My experience with this has been that it overwhelms the team doing reviews and leads to slow turn around times.

Possible, sure. Although I see Bioconda less susceptible to that than conda-forge since we have that unified repository here. So the probability someone sees/reviews a PR should be higher compared to single feedstocks on conda-forge. But thanks for sharing that concern!

@tiagoantao
Copy link
Member
tiagoantao commented Jan 17, 2018 via email

@epruesse
Copy link
Member

Let's start by having a review team and seeing how long it takes for reviews to be done in the worst case. Turnaround time and stability are competing interests, and that discussion is out of scope here.

That said... I'm using Bioconda to provision the tools I need in "production". In that environment I can easily live with a week's turnaround time, but breakage (networkx...grrr) is a problem. IIRC Bioconda policy is releases only, not pre-releases or alphas, for that exact reason. The only time I really need something to get uploaded quickly is to fix breakage, which I'd much prefer not to have happened in the first place. Hence: +1 for mandatory reviews.

@bgruening
Copy link
Member
bgruening commented Jan 17, 2018

Lets move the discussion about mandatory review into a new thread (@epruesse, @mbargull do you want to do this?).

@mbargull #7293 :) and I added you to a few groups.

I created also a review team and added @mbargull, @epruesse and myself to it.
There is also now a build-system team for problems with travis and bioconda-utils.

@druvus
Copy link
Member
druvus commented Jan 17, 2018

I am happy to be added to review and osx.

@bgruening
Copy link
Member

@druvus added you thanks a lot!

@mbargull
Copy link
Member

Lets more the discussion about mandatory review into a new thread (@epruesse, @mbargull do you want to do this?).

Good idea, done as #7294.

@mbargull #7293 :) and I added you to a few groups.

I created also a review team and added @mbargull, @epruesse and myself to it.
There is also now a build-system team for problems with travis and bioconda-utils.

Thank you very much!

@cbrueffer
Copy link
Member

I agree that a review team would be useful, feel free to add me.

@bgruening
Copy link
Member

Done! :) Thanks @cbrueffer!

@rvalieris
Copy link
Member

I would like to be added to review, perl and R.

@bgruening
Copy link
Member

@rvalieris done! Thanks!

@souravsingh
Copy link
Member

I would like to be added to review and R teams.

@mbargull
Copy link
Member

FYI for everyone: You can also hit the "Request to join" button on any existing team's member page, e.g., https://github.com/orgs/bioconda/teams/review/members, if you would like to join that team.

@bgruening
Copy link
Member

@souravsingh added you!

@bgruening
Copy link
Member

We do have those teams now and they are super useful. conda-forge also has them so I think we can close this issue. The only thing that is missing is that people make use of this regularly. I have the impression this is under-used :)

But this is an other problem :)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests