Vote on integration of Yarn.

In the TSC meeting today, those present agreed that we should call for a vote. This issue is to discuss with the full TSC since not all members were in attendance.

The issue driving the discussion is: deps: add Yarn 1.22.5 #37277 and the related issue is: nodejs/node#35398

You can catch up on the discussion in the minutes: #1011

My first cut at the options for the vote based on the discussion in meeting are:

we should integrate yarn/land 37277
we should integrate corepack/land 35398
my preference is that we integrate yarn, but if that does not get enough votes we should integrate corepack
my preference is that we integrate corepack, but if that does not get enough votes we should land yarn
I don't think we should land yarn or corepack
I don't think we should land yarn or corepack because there is not enough information to make a good decision yet

The last 2 will both be counted together but would provide insight if a no vote is due to an objection or a lack of information (since it has been raise that more info may be necessary to make a good decision).

@nodejs/tsc please review and indicate if there are any objections to move to a vote, and suggestions for improvements/changes to the options suggested for the vote.

These options look good. I'm +1 on moving to vote

Moving to a vote and the set of options look good to me, though "land ABC" imo should be replaced with "target landing ABC", so that e.g. voting for 1-4 shouldn't imply an approval of the current state of those PRs without comments or be treated as a code review.

Also, perhaps an explicit abstain option could be useful.

+1 on moving to vote

Would this issue be suitable for a Condorcet? That would allow TSC members to express more precision what they want for Node.js, and it's probably more "fair" than having to chose only one option. For a Condorcet vote, I think the options could be (in alphabetical order):

Keep npm, integrate Corepack (not Yarn).
Keep npm, integrate Yarn (not Corepack).
Keep npm, integrate Yarn and Corepack.
Keep npm, do not integrate Corepack nor Yarn.
Remove npm, do not integrate Corepack nor Yarn.
Remove npm, integrate Corepack (let npm and Yarn decide if they want to be in Corepack).
Remove npm, integrate Yarn (do not integrate Corepack).

Is “remove npm” even on the table?

remove npm was not on the table during the last TSC meeting.

Is “remove npm” even on the table?

Maybe I'm not wording it correctly, I mean stop bundling npm by default, not completely removing it of course. I remember some participants of last TSC meeting stating they'd rather do that than bundling yarn. IMHO it makes more sense to discuss the status of npm in Node.js along side yarn integration, but please disregard if that's off-topic.

My point was to suggest using Condorcet for this vote, not removing npm.

+1 on condorcet.

Condorcet is not a single method but a group of similar methods.

More specifically: given N options, start with ranking each option on a relative preference scale (1-N, allowing duplicates).
Then use some of the methods to rank those (those are different)

While there are methods that e.g. disallow duplicate rankings (simple ranking) in the voting process, there seems to be little benefits of that and that just complicates the decision process.

I'm wondering if Condorcet is going to let TSE members better express their opinion? It's going to take more work to set up/etc. so want to be sure pushing out the vote to do that is beneficial. In particular I'm not sure how it handles the "but only if X" does not win.

@ChALkeR

@ChALkeR did you come across the use of Condorcet for non-elections ? I see most articles reference electing people versus making a decision?

did you come across the use of Condorcet for non-elections ? I see most articles reference electing people versus making a decision?

I've seen used to pick a name. It's just a voting system, I don't think there is any difference if it's used for electing people vs choosing ideas.

I'm wondering if Condorcet is going to let TSE members better express their opinion? It's going to take more work to set up/etc. so want to be sure pushing out the vote to do that is beneficial. In particular I'm not sure how it handles the "but only if X" does not win.

From what I've read, it has some significant advantages including allow voter to express their opinion (instead of going all in for only 1 proposition, you rank them all, so you can better express where you stand on the issue), and you can add "but only if X" propositions to the list without splitting the votes. I reckon it does take more work to set it up, let me know if I can help, I think it would be useful to have some tooling to assist with that if we haven't a Condorcet voting system already (even if we end up not using it for this particular vote).

Quick question: are the newly nominated folks already onboarded or are we waiting on anything? Will they participate in the vote? If they are not, should we wait until they are onboarded before voting?

Preview page for the vote (vote is not open yet, actual link for the vote will be sent directly to TSC members emails):

I tried to keep the summary short, but we can write something more elaborate if folks think it'll be helpful.

The setting I'm using are listed below. Please let me know if there's anything folks thing should be changed.

@Trott

@Trott

I think there are diminishing returns to introducing more nuanced ranked voting than RCV.

Offtopic on meta

To me, RCV looks more complex that simple scoring, because it entangles options between each other. I would have preferred something like this instead:

Which provides the data that we need, and for me looks easier to both use as a data source (reasons above) and to fill in, because it avoids entangled options, i.e. comparing them to each other, and instead asks a simpler more grounded question.

That said, I'm ok with RCV for the sake of continuing this, perhaps we could move the meta discussion (for future) to some other channel from this PR. I hid that under a spoiler, hope that's ok.

Vote link sent to @nodejs/tsc members via email listed on https://github.com/nodejs/node#tsc-technical-steering-committee. If you didn't receive an email, please let me know and I'll send you a code to vote. Being a free vote session on OpaVote means we have 7 days to finish the vote, so please vote ASAP. If you are going to abstain, please vote without selecting any options so we know you're aware of the vote.

I'll try to send daily emails to members who didn't vote yet to make sure everyone has a chance to vote.

Daily update: 14 voted so far, 8 remaining.

Daily update: 17 voted, 5 remaining.

This is hopefully obvious and may already be covered above somewhere that I'm not seeing in a quick skim, but I want to make sure we all agree that a vote for including corepack or bundling yarn is not a vote for any particular implementation. For example, hypothetically, I may be in favor of shipping corepack, but only if we can do it in a way that does not change anything for existing npm users--in other words, it can't introduce any additional friction/burden for existing users and CI automation and whatever else. So I might be in favor of corepack but opposed to a specific PR that adds it.

Daily update: 20 voted, 2 remaining.

Daily update: 21 voted, 1 remaining.

@Trott

@Trott to confirm from the minutes of the last TSC meeting:

Vote is on general direction, not land as is, PR’s need work/review/and may be experimental

Still 1 remaining. I sent a final reminder to them and will close the vote in 24 hours. Unfortunately I don't know if their email is up to date on our mailing list (see #1022) and I don't want to ping them directly here without their consent (but I also can't reach them anywhere else). So I'm going to cc @nodejs/tsc once more just in case

Vote is closed. Winner is "add corepack" with 63.2% of votes as first choice. Ballouts have been shared with TSC members in private.

OpaVote Election Results (https://www.OpaVote.com/)

Vote on integration of Yarn

There are 3 candidates competing for 1 seats. The number of voters is 21 and
there were 19 valid votes and 2 empty votes.

Counting votes using San Francisco RCV.

 R|Add corepack      |Add yarn          |Leave things as is|Exhausted         
  |                  |                  |                  |                  
==============================================================================
 1|                12|                 2|                 5|                 0
  |---------------------------------------------------------------------------
  | Count of first choices.
==============================================================================
 2|                18|                  |                  |                 1
  |---------------------------------------------------------------------------
  | Count after eliminating Add yarn and Leave things as is and transferring
  | votes. All losing candidates are eliminated. Candidate Add corepack is
  | elected.

Winner is Add corepack.

This means we're not moving forward with #37277. We can revisit nodejs/node#35398 to see if there are any technical concerns left before landing.

@mmarchini

@mmarchini thanks for running the vote and reporting the results.

Commented on nodejs/node#37277 and closed it. Closing this issue as the vote is complete.

Navigation Menu

Repository navigation

Vote on integration of Yarn. #1012

23 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Vote on integration of Yarn. #1012

Description

Activity

jasnell commented on Apr 9, 2021

ChALkeR commented on Apr 9, 2021

ChALkeR commented on Apr 9, 2021

gireeshpunathil commented on Apr 10, 2021

aduh95 commented on Apr 10, 2021

ljharb commented on Apr 10, 2021

mcollina commented on Apr 10, 2021

aduh95 commented on Apr 10, 2021

mcollina commented on Apr 10, 2021

DerekNonGeneric commented on Apr 11, 2021

ChALkeR commented on Apr 11, 2021

nschonni commented on Apr 11, 2021

mhdawson commented on Apr 12, 2021

mhdawson commented on Apr 12, 2021

aduh95 commented on Apr 12, 2021

23 remaining items

mmarchini commented on Apr 15, 2021

mmarchini commented on Apr 15, 2021

ChALkeR commented on Apr 15, 2021

mmarchini commented on Apr 22, 2021

mmarchini commented on Apr 23, 2021

mmarchini commented on Apr 24, 2021

Trott commented on Apr 25, 2021

mmarchini commented on Apr 25, 2021

mmarchini commented on Apr 26, 2021

mhdawson commented on Apr 26, 2021

mmarchini commented on Apr 27, 2021

mmarchini commented on Apr 28, 2021

mhdawson commented on Apr 28, 2021

mhdawson commented on Apr 29, 2021

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions