Who controls glibc? [LWN.net]

Who controls glibc?

Posted May 7, 2018 19:20 UTC (Mon) by siddhesh (guest, #64914)
In reply to: Who controls glibc? by ami
Parent article: Who controls glibc?

We (the glibc developers/maintainers) have made it pretty clear what we prefer, please read the thread.

to post comments

Who controls glibc?

Posted May 7, 2018 19:37 UTC (Mon) by ami (guest, #5280) [Link] (20 responses)

I've read bits of it. How do you determine the consensus from that?

Say you have 50 developers. 1 is strongly in favor of a position, 4 are strongly opposed, and 45 really don't care. If you read a discussion thread about this, you might conclude that the consensus is opposed, but the real consensus is closer to "meh".

Anyway, I'm not asking to be a jerk. I genuinely curious. Do glibc developers have some sort of polling mechanism to determine consensus?

Who controls glibc?

Posted May 7, 2018 19:56 UTC (Mon) by siddhesh (guest, #64914) [Link] (19 responses)

Consensus is different from unanimity, so you don't necessarily need everyone to agree exactly on a decision. You only need to be sure that there is no strong sustained opposition. To give a technical example, I may not like a hackish patch but don't hold it up on the condition that the submitter posts a follow up to clean it up. Conversely, I may decide to oppose inclusion of a patch because it adds quadratic behaviour to a case I care about but others may not. In such cases the community negotiates till we reach a resolution.

When 45 out of 50 do not respond, it's not necessary they they don't care, quite often the case is that they agree with the current state, which may either be of consensus or of a sustained opposition and don't want to get involved in the conversation unless absolutely necessary. Most contributors to glibc (even the most active ones) are usually working on the project in addition to another project, like gcc or the kernel so they're juggling multiple things at once.

The article links to a wiki page that describes the consensus rules we use in glibc in quite a bit of detail.

Who controls glibc?

Posted May 7, 2018 22:15 UTC (Mon) by k8to (guest, #15413) [Link] (8 responses)

A bit off topic, but my 2 cents about consensus:

In the Quaker tradition of consensus decision making, it's a matter of making a best effort to find what the group as a whole wants.

In that tradition it's considered a mistake to try to build a consensus without hearing from at least most of the members. However, it's also built around the idea that people who are going to participate essentially opt into discussion in certain areas, and have explicitly taken on the duty of participating, even when it's time consuming and slow, which are par for the course in classic consensus decision making. Essentially, everyone is given time to speak about what they think and what they feel, and when the turn comes to you, you can't just pass or stay silent. It would be difficult to adapt this fully to an online medium. Functioning well, this can include outcomes such as participants deciding that they really are too far outside the group needs/wants to be part of the decision and essentially pulling up their stake. It can also include the group as a whole working hard to understand the views and opinions of one who sees it quite differently.

In the general modern interpretation of consensus decision making, participants are often far less committed to fully participate, and would rather simply ignore issues that are insufficiently important to them, or have a poor ratio of importance to expected annoyance. In such an environment, I really don't see what you can do besides try to remember that each participant's views should be given real consideration, and at the same time that it is important and necessary to find the outcome/view/decision that best represents the group overall. Even that is hard to instill in a free assocation of developers, and I think it's generally not perceived that this mode of thinking is important and possibly essential.

In short, consensus is hard. We're not really trained to do it. Most of us don't really fully appreciate the components to achieve it. Expecting strong consensus on mailing lists is perhaps a foolish pursuit. But considering how things are playing out here, I do wish more of the participants valued those principles.

Who controls glibc?

Posted May 8, 2018 18:02 UTC (Tue) by lamby (subscriber, #42621) [Link] (2 responses)

Where can I learn more about the Quaker model vs. the "modern" model?

Who controls glibc?

Posted May 9, 2018 3:23 UTC (Wed) by k8to (guest, #15413) [Link]

I don't know. It's just based on personal experiences in my life of what people perceive the word to mean among differing groups.

Who controls glibc?

Posted May 9, 2018 11:28 UTC (Wed) by TheIllustriousYou (guest, #124302) [Link]

If you have a Quaker meetinghouse near you, you might pay a visit and ask them. I know a few; I'll ask if any can help you learn more.

Who controls glibc?

Posted May 8, 2018 18:47 UTC (Tue) by ClaudeRubinson (subscriber, #11921) [Link] (4 responses)

The Quaker decision-making process seeks to discern God's will (not the group's). It is this belief that a truth exists outside of the decision-making body that distinguishes it from conventional (secular) decision-making processes, including consensus and voting.

Who controls glibc?

Posted May 9, 2018 3:22 UTC (Wed) by k8to (guest, #15413) [Link] (3 responses)

This is slightly hair-splitting, because Friends believe there is that of good within people. In practice, there are many decisions made within a monthly meeting which are not viewed by most people as having a significant spiritual dimension.

Who controls glibc?

Posted May 9, 2018 3:22 UTC (Wed) by k8to (guest, #15413) [Link]

Err, "that of god".

Who controls glibc?

Posted May 9, 2018 16:53 UTC (Wed) by ClaudeRubinson (subscriber, #11921) [Link] (1 responses)

Quaker process seeks to suppress individual egos and preferences in search of an external truth that exists beyond and apart from the individuals involved. Secular consensus making, on the other hand, seeks to accommodate and encompass the different preferences, viewpoints, etc by asking "What can we all agree on?" This is precisely what makes arriving at consensus so challenging. (The different goals is also one reason why Quaker decision-making practices generally don't work well in a secular context.)

Who controls glibc?

Posted May 9, 2018 17:32 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link]

Entirely correct. My religion tells me that the only true way to consensus is personal combat with edged or blunt weapons.

It's entirely unbiased and fair - there's no nonsense with differing interpretations of "scripture" or arguments about its application.

You either win and vanquish your foes or lose - it's absolute morality, imposed by the all-seeing eyes of gods. After all, they won't let you lose if you are correct, otherwise they won't be omnipotent and all-seeing.

I think that this kind of decision process should be universal.

Who controls glibc?

Posted May 8, 2018 13:01 UTC (Tue) by yshuiv7 (guest, #96631) [Link] (9 responses)

Wait a second. You are saying you can assume someone is agreeing with you, if they don't respond?

I think there might be some problem with this assumption.

Who controls glibc?

Posted May 8, 2018 13:14 UTC (Tue) by siddhesh (guest, #64914) [Link]

No it's not as simple as that. Depending on the nature of the change, you as maintainer are trusted to decide how long you'd like to wait for agreement and how many people you'd like to review the change before committing it. Every once in a while you might get it wrong but that's not the end because we can always discuss to gather consensus either way.

Again, I'm simplifying, please read the consensus wiki page (linked in the article) which is way more exhaustive.

Who controls glibc?

Posted May 8, 2018 18:28 UTC (Tue) by sjfriedl (✭ supporter ✭, #10111) [Link]

It's not fair to read no response as "agreement", but it is fair to read it as "not opposed"

Who controls glibc?

Posted May 10, 2018 3:08 UTC (Thu) by bferrell (subscriber, #624) [Link] (1 responses)

> Wait a second. You are saying you can assume someone is agreeing with you, if they don't respond?
> I think there might be some problem with this assumption.

That's actually the concept behind how abstentions are counted... If you abstain, it's not a disagreement, It's counted as if in favor.

At least in areas I've seen it used, which kind of stinks because people thing it means "I disagree, but refuse to vote"

Who controls glibc?

Posted May 10, 2018 9:10 UTC (Thu) by karkhaz (subscriber, #99844) [Link]

Systems that have a compulsory voting system typically have abstention as an explicit option, which allows for the kind of protest you mentioned: it is compulsory to cast a vote, but you may cast a vote for "none of the above".

This is used on a national elections level in Australia, where voting is compulsory. There isn't actually a "none of the above" box but it's perfectly legal to turn up to the polling station and hand in a ruined ballot card, which is the defacto way of abstaining.

It's indeed a bit presumptuous to lump "I don't care" with "I strongly disagree with all options" in the absence of a voting system that doesn't distinguish between those options. However, on a software project mailing list, it's not the case that there are a fixed number of options to vote for---if somebody disagrees with all possible proposals that have already been suggested, they're free to introduce an alternative suggestion. So in that context it does seem reasonable to read an abstention as "I don't care about this issue" or "I agree with whatever the current consensus is".

Who controls glibc?

Posted May 10, 2018 8:53 UTC (Thu) by zenaan (guest, #3778) [Link] (4 responses)

> Wait a second. You are saying you can assume someone is agreeing with you, if they don't respond?

It's called "tacit agreement" - yes, it's agreement by default - it is an effective (whether intentional or not) surrender to the "vocal majority" even though that vocal majority may be an absolute or relative minority.

But, in a public forum (or "modern democracy") tacit consent also implies a high likelihood of the tyranny of the minority - i.e. those who squawk the loudest get to "win" (or get the government benefits, etc).

Those who seek "safe spaces" (whatever the hell they are) are also called cry-bullies.

The SJWs or "social justice warriors" who actively "create safe spaces" by for example (to pick a totally random example) removing a short yet politically incorrect joke from documentation, are often engaged in some level of "white knighting" or virtue signaling and in any case are subsidizing bad behaviour.

Now, what could also arguably be alleged as the "bad behaviour" of a politically incorrect joke, is also a healthy poke in the ribs of the cry bullies, those trigger-ready snowflakes who self proclaim to melt at the sight of any of a million trigger words - and by publishing their melting quality far and wide, publicly and loudly, they are bullying the rest of us.

We owe it to our dignity to NOT surrender our freedom of speech or our "freedom to trigger" in our public, work and play spaces, to the dictates of the covert passive aggressive cry bullies.

And remember folks, EVERYTHING we say, all our speech, is effectively political in some way. Those who remove a short yet politically-incorrect joke from documentation are making a political statement, doing a political action, saying "this is a safe space for cry bullies, you will not pollute our safe space with politically incorrect jokes", even though it is (usually) in the guise of "we want no politics in our docs".

Who controls glibc?

Posted May 12, 2018 14:12 UTC (Sat) by niner (guest, #26151) [Link] (2 responses)

Just in case you don't know: insulting your opponent usually does not help you win an argument and will not convince anyone. All it does is reflect badly on you and reduce your credibility.

Who controls glibc?

Posted May 12, 2018 14:49 UTC (Sat) by mpr22 (subscriber, #60784) [Link] (1 responses)

I submit that someone who unironically links to Zero Hedge doesn't care about credibility in the first place :)

Who controls glibc?

Posted Nov 8, 2018 16:39 UTC (Thu) by deepfire (guest, #26138) [Link]

To clarify, that statement of yours says far more about your opinion of ZH, rather than about the poster.

Who controls glibc?

Posted May 16, 2018 21:09 UTC (Wed) by tartley (subscriber, #96301) [Link]

No bullying occurred at all. A suggestion was made, discussed, and eventually acted on despite some opposition. This is the normal process.

No censorship occurred at all. RMS or anyone else are free to continue making whatever jokes they want, or can publish a fork of the manual with the joke intact.

The "safe spaces" you complain about only require a moderate level of consideration for others.

I absolutely concede that sometimes people abuse that mechanism, petitioning to impose unreasonable burdens on others. In such cases, your objections would have some merit. But that has not remotely happened in this case, and unless it does, being respectful of other people's wishes is a very low bar indeed.

The joke's a bit crass. Some people don't like it. Most people don't care. Delete it and move on. No harm results. Objections to this are just crying wolf. Save it for an instance that genuinely needs attention.