I'm a big fan of RMS, but I've got to say "No". He's not a maintainer. He doesn't get a vote. If he wants to work his way back into that project, then I've got no problem with it. But he doesn't. It's like the CEO telling you that he wants a joke in the technical documentation for one of your libraries. They might get away with it because of their position, but it's an abuse of that position even if the CEO used to be coder #1 on the project.

Not that it doesn't happen all over the place...

If there is business justification of joke in documentation then maintainer has no authority to say no.

RMS is a CEO and he ultimately controls the project goals. If you think otherwise you fork.

Is there really a clear line of authority between the glibc maintainers and RMS? It certainly doesn't seem like it. Authority decays, and RMS's is gone, over glibc at least. Compare with GvR: Guido is involved day-to-day, talks to people, makes decisions. No one questions whether he has power, because he uses it, although they certainly question the way he uses it.

I think the monarchy of the United Kingdom is good way to look at RMS's position here. On paper the monarch has a lot of power, but the reality is that they don't and trying to use it would provoke exactly the kind of crisis we're seeing here.

I have a hard time imagining the monarch insisting on a joke. Not just Queen Elizabeth, but any hereditary monarch from the modern age. Monarchs retain authority because they're good at politics... RMS, not so much. That's not to say his political positions are wrong, just that the man is impolitic at times.

Monarchs exert soft power pretty much constantly, with public appearances, knighthoods, and recognition of great deeds. They often end up as the most popular political figures of their republics. So while monarchs are a fossil species in modern governments (since WWI), they retain enough of the royal prerogative to exercise those paper rights it when it actually matters. I think we're all second-guessing RMS here because he's using the perogative in a situation where it truly doesn't matter.

Case in point: In Japan Prime Minister Abe wanted to change the constitution to allow Japan to engage in external conflicts if allies were in danger. The emperor of Japan suddenly said, "I'm old, you know. I think I want to retire, but there is no constitutional law that will allow me to abdicate. Please write this law ASAP". Prime Minister Abe does not have resources to do both constitutional changes at the same time...

The emperor of Japan is really awesome.

I guess they learned from the last few times when military juntas took control of Japan and burned everything to the ground, in the end.

> Monarchs [...] often end up as the most popular political figures of their republics.

I can't quite put my finger on it, but I think there's something wrong with that statement :p

I'm going to... stick my neck out... and venture that republics have usually got rid of their monarchs.

Har har, yeah, point taken, but many constitutional monarchies are de facto republics. The UK, Sweden, the Netherlands, Japan, Norway. I was trying to distinguish from the hereditary monarchies that aren't republican, like the People's Democratic Republic of Korea.

> Har har, yeah, point taken, but many constitutional monarchies are de facto republics.

No, they are just representative democracies. Which aren't coextensive with republics: you can be either without being the other.

> Is there really a clear line of authority between the glibc maintainers and RMS?

Obviously, yes. RMS is currently the president of the FSF and glibc is a FSF project, created and directed and maintained and funded by the FSF.

Just because some developer parachutes into a project that doesn't mean the project owners lose their rights over the project, or that the developer becomes the new owner.

If you want to contribute to a project, you do so by respecting and complying with the wishes of those directing the policy. The FSF started the glibc project and manages glibc's development and establishes the policies followed by the glibc project. It's the FSF, and not any random developer who just so happena to have commit access to a repository.

That's not the way a company works. A CEO is not dictator, and there are more paths to resolution than simply following orders or quitting.

Why then should the dictatorial mode be the way a non-profit or volunteer effort works?

Moreover, in a company it's the Board of Directors which delegates authority to the CEO. And the shareholders could override even them.

So if we interpret glibc as as a cooperative where the shareholders are also the developers and others working on the project, then certainly the shareholders could override management.

>>> A CEO is not dictator

well, he definitely is for those under him (not those besides).

I've been trying to figure out an answer which was more concrete than "I disagree", and ran into the problem that I can't figure out what "dictator" is supposed to mean, other than "person in authority whom I disagree with."

I mean, people called Obama a dictator (https://www.marketwatch.com/story/americans-newly-see-obama-... ), Jeremy Corby (http://www.livetradingnews.com/jeremy-corbyn-turns-dictator-... ), a teacher's union (https://www.huffingtonpost.com/dave-stieber/i-am-a-dictator-... ), and more.

When I wrote that, I was thinking specifically of how a union could override a CEO's demands demands, eg, by (threat of) lawsuit over breaking an employment agreement, or by collective action.

Then again, a dictator also has to worry about a mass uprising.

Then again again, so does a president - the US president is the chief of the executive branch of the US.

Do you think a CEO is dictator by definition, or that there is no way for a CEO to be anything other than a dictator? If so, what do you mean by dictator?

I think there is no other way to be a dictator (as long as he's appointed byt the board).

Now what is a dictator to me ? A dictator is someone you can't disagree with. The dictator may be benevolent or not, may exercise his power or not, but in the end he's the one in power. And you can't disagree with him because you have much less power (he can fire you, he's got the money, he decides on what you work, you're the one who needs the job, etc.). Just try to contest the authority (not the ideas, the style or whatever, just the authority) of your boss, and you'll see what happens.

Now from the board of director point of view, the CEO is not a dictator, he's someone who's running the company and giving advice on how to make it grow, make it better, whatever. So he's definitely in a positive role.

Could he be something else than a dictator ? Maybe. He could be "secretary" and then the power could be given to a part of the company's personnel, like a union. Those people would attend the board of director. But that would assume the fact that the board is willing to do that... I guess it's easier to scream on the CEO than on a group of people :-)

The Roman office of dictator (which is the origin of the term) was elected with absolute power in order to deal with a crisis -- generally speaking, a major war, and that usually involving an invasion of Roman territory by hostile forces. The term of office was limited to one year, though dictators were sometimes re-elected.

Roman dictator terms were generally either 6 months or until the task they were appointed to complete was finished. The Senate was capable of extending this term or re-electing the dictator depending on the circumstances.

You are right, I was relying on memory.

Thank you for your clarification.

I found an example of a CEO who fired an at-will employee and as a result the company was successfully sued for breaking an oral contract where the board of directors had promised that the employee would not be fired for speaking about workplace problems. https://www.jacksonlewis.com/publication/south-carolina-jury...

That would be an example of where the CEO is not a dictator, yes?

In some states, a CEO cannot fire someone without just cause, or for reasons motivated by malice or made in bad faith. And there are civil rights, disability rights, etc. which also prevent certain CEO actions. But these can also be viewed as dictatorial government actions overriding dictatorial corporate actions. :)

That's why I brought up union representation and employee contract rights which prohibit the CEO from certain actions.

I think there are also cases where the CEO is unable to impose demands because certain key staff would ignore them or leave the job, or the CEO fears the public reaction should that happen. Eg, http://tvtropes.org/pmwiki/pmwiki.php/Main/UltimateJobSecuri... .

I think it's important to distinguish between limitations on power imposed after the fact, and limitations on power that prevent one from exercising their power. CEOs are not dictators in that they answer to a board, shareholders and, ultimately, the law. But those limits tend not to prevent CEOs from exercising their power, but are rather consequences of exercising their power. On a particular day, when CEOs say something should be done (such as firing a particular person or saying something publicly) it tends to happen.

Whether or not we call a CEO a "dictator" is perhaps a distraction - although I think the word usually fits in the colloquial way people tend to use it. The important thing, I think, is the recognition that CEOs orders tend to be followed, sometimes even against the desires of people executing those orders. The controls placed on CEO's power tend to be consequences rather than immediate limitations.

People call all sorts of things, all sorts of words. It's common in discourse to use a little hyperbole. That does not make Obama, or Jeremy Corbyn, bona fide dictators.

Even as stated, you miss the mark. The president is chief of the executive branch of the US, yes. But the president is not chief of congress, or the judicial system.

Certainly. I was the one who wrote that "the CEO is not a dictator".

This is trivially true as the CEO is not the authoritarian head of a country.

So I must have meant something a little hyperbolic.

I just don't know how to define that in any meaningful sense so as to be able to talk about it.

I don't know what mark I missed. I objected to the llukas' proposition that the CEO in a business "ultimately controls the project goals." I agree that in the US government, the chief executive does not have full control. I think it's meaningful point that the chief executive does not necessarily have final control.

Personally, I think it's a simpler story to talk about the CEO as some sort of heroic authoritarian leader, which is why we end up with those sorts of stories.

They should. Glibc breaks its interface every few years. I’d love to see all the GNU tools migrate over to the Linux foundation, where they would be stable, safe, and free from autocracy.

Well, let's not forget that "CEO" wrote that technical documentation. And probably also wrote (at least an older version of) the code, too.

Which counts for absolutely nothing. A CEO running around the floor of the business telling programmers what jokes belong in what docs is a shit CEO no matter what he didn’t years past.

I fail to see why, if that joke is actually a clear statement of the fundamental message that the project was designed to convey, and the whole reason why the FSF exists.

not quite true,

https://en.wikipedia.org/wiki/GNU_C_Library#History

He might not be the maintainer, but he is the BDFL. It's about respect. See what happened to Drepper who violated his respect.

I like the joke, and I sympathize with RMS to let it in. In any case he opposed it so there is now a power struggle about the control of his project. I would love to see a better maintained glibc btw. Too much politics, not enough competence.

The dictator in question certainly doesn't seem Benevolent to me in this case.

> The dictator in question certainly doesn't seem Benevolent to me in this case.

I don't see the point. RMS directs the organization and establises the policy it follows. I fail to see how a maintainer could feel entitled to dictate that policy just because he has access to the repository. That would be like a printing press janitor feeling entitled to censor what he feels a newspaper should not publish.

It's not like he is not benevolent either.

He certainly is. He tries to protect the project from morons, which invaded most bigger projects which are not cultured in technical discussions anymore.

Compare that to the education of doctors. They need to persist on proper terminology and are hardened in a long process. PC is absolutely counterproductive there.

Many obviously did not get the reference in this paragraph, but it still relevant, as this episode showed. Abort vs kill is loaded by a long fight, hence the reference about future censorship attempts.

> PC is absolutely counterproductive there

A. No, the terminology used by medical professionals is constantly shifting to do things like acknowledge that people with issues are still people.

B. Your reasoning is presumably about confusion and lack of clarity, yes? How exactly is a reference to a law that has never been proposed serving that?

Well see what Drepper did. CTRL-f for "Stallman" in these glibc release notes:

https://sourceware.org/ml/libc-announce/2001/msg00000.html

While I agree with Ulrich, it is pretty brutal.

Yes, that was what I was referring to. It backfired big time.

See what Drepper is doing now: https://www.linkedin.com/in/ulrichdrepper He went from Technical Director at Redhat (i.e. glibc maintainer), to VP at Goldman Sachs (i.e. booted and being called an asshole), and now he is Engineer at Redhat.

Glibc spent over a decade under Ulrich Drepper, who many viewed as a dictator. This resulted in forks and major linux distributions no longer using glibc. About 6 years ago, Drepper was removed and glibc moved to "cooperative community development"[0].

The community spoke and said they did not like dictators in development of glibc. Since Drepper's removal, many would agree that the quality of glibc has increased significantly. RMS is well aware of this and the delicate political situation. He should recognize that coming in and declaring himself in charge would not play well with a community that stated about their maintainers following Drepper: "This does not confer any extra ability to make decisions for the project; community consensus is what matters there."

[0] https://sourceware.org/ml/libc-alpha/2012-03/msg01040.html

I beg to disagree. The quality post-Drepper certainly did not improve. Bugs are getting worse, needed features are still not implemented, needed performance improvements are sidelined (bsd libc still 20% faster with modern CPUs), crazy but useless ASM optimizations are being added, where e.g. proper loops written in C gain much more. Just not on gcc.

Same as with stdlib++ btw, where strings are now uselessly bloated and most data structures are getting useless also.

Where are utf8 strings? u8"" constants only is for nothing. Everybody needs an u8 api.

Where are the secure _s extensions? Pickering about callbacks and Microsoft is all they can do. But they rather add env hooks. __bos/alloc_size support is getting better, but gcc still cannot do proper usable const expressions in C. It's a nightmare.

You still cannot compare strings, even not with the crazy wchar_t strings (size 2 or 4), nobody uses anymore. wcscmp only compares buffers, but not strings according to the unicode rules. I believe strings would be pretty important to support, not?

What about a fast malloc? Not even ptmalloc3 went in, there's still stone-age ptmalloc2.

double-close leads to crashes, really? Similar to freopen with NULL.

This is your great community consensus. BSD or even musl are far ahead.

> (bsd libc still 20% faster with modern CPUs

You have a source for that? it wouldn't surprise me, GNU code tends to be more bloated than a similar BSD licensed project.

My own benchmarks. The trick is SSE support. darwin/bsd has that builtin, esp. since darwin can guarantee that their chips do support SSE. Generic linux builds do not, and -march-native or switching to sse optimized shared libs is rarely used.

But recently also bos and align_size support got better in other libc's, which do use clang and not gcc. gcc sucks big time with those optimizations. freebsd and darwin all use clang. This is in the ~60% ball figure.

Every 64-bit OS can guarantee SSE2 support, because it's baked into the x86-64 spec. Every single 64-bit Linux build can and does use SSE2 (unless someone explicitly turned that off for no reason). If there are performance differences, it's not due to chip support and crippled builds for compatibility.

oops: s,align_size,alloc_size,

The size of dynamically allocated malloc'ed structures, not just constants. The majority of pointers have an alloc_size, but no object_size (i.e. bos: __builtin_object_size).

> Lets think how many hours RMS has put into GNU over the years. That should count for something, no?

In terms of personal respect, sure.

In terms of attention given to an argument, maybe, but that's also how you end up with a narrow bubble.

In terms of retaining a unclear joke designed as backhanded political advocacy on an issue that isn't central to the shared ideology that defines GNU in the technical documentation of a GNU project? I can't see it, no.

I thought the whole point of Free Software idea is that contributing to a software project does not make you its exclusive owner (after you released it under Free Software license) and its users subject to your whims. It is ironic that at the top of this project is a person who would very much like to subject the project to his whims, and for quite a trivial reason at that.

No. Free software is about respecting the four freedoms not about what you say above. If it counts to anything, you get the choice to fork the project at any time.

Count for what? Should it excuse his behavior? Should we excuse the behavior of famous and notable people because of their accomplishments?

I don't think it's a matter of excusing, it's whether he has ultimate authority based on the fact that he created glibc and he's the head of GNU. Talking about excusing assumes he did something wrong, which is jumping the gun, since that's exactly what's being discussed.

I have a hard time seeing where he did something right here. He is behaving like a petulant child because people didn't find his joke funny two decades later.

He is the "boss". No doubt about it, he is the copyright holder, everyone working on the project has surrendered copyright over to him. I've got my strong reservations on BDFLs and the way their egos get stroked, but it is what it is.

But he is not a "leader". Because a leader does not cut the legs out from under the people he delegates to run things. He has every right to insist that GNU maintains his dumb joke... and everyone who's put in all of the hard work and effort into the project over these more recent years has the absolute right to leave over it.

Oh, I don't disagree. I just don't think we have anything to excuse, that's between the maintainers and him.

If RMS wasn't the project lead, but any other maintainer, then his behavior of making such a big fuss about this particular thing (which, very importantly, _is not a very important thing!_), against the consensus of many other maintainers, is something.

If he were on the same level as other people, this behavior wouldn't be accepted. So the question then becomes:

Does the fact that he's "the boss" excuse his behavior of fighting against the consensus on an issue of this level of importance?

Alright, so where is that line drawn? RMS is the project lead, but it's a community project and at some point the consensus will be that he is just wrong. Clearly that line must exist somewhere, or eglibc would never have been. But the existence of a fork isn't the minimum point since that takes quite a lot to overcome the inertia of forming a new project.

The point I keep coming back to is: What is the purpose of the documentation? What goal does it achieve?

What is the point of the project? What goal does it achieve?

There was a gcc frontend/backend clean architecture dispute: some wanted it so that gcc could be used for more things that clang is being used for. RMS argued against it, on the basis that GCC technical architecture should deliberately avoid it working with proprietary non-GPL backends.

The official story is that a technically worse compiler than loses market share is still superior for the FSF goals of pushing free software: they are willing to accept people moving to clang to avoid GCC from being the front half of a non-free compiler.

That goal is inherently political. It isn't obvious whether drawing attention to harmful suppression of information by the government is also in the projects mandate, but it is obvious that the projects mandate isn't just to make the best technical and professional result.

https://gcc.gnu.org/ml/gcc/2014-01/msg00209.html

https://gcc.gnu.org/ml/gcc/2014-01/msg00178.html

People sometimes overlook the fact that the FSF is clearly a political organization, the GNU project places politics over technology, and Stallman himself is a political activist first and foremost. Yet people forget this and treat GNU stuff like technology-driven projects.

Personally, I'm much more comfortable with the approach Linus Torvalds takes, where writing good, functional software takes precedence over purity concerns, and where technical limitations would not be introduced with the purpose of complicating interactions with proprietary software.

Those arguments don't work.

Just because a goal of the project is an inherently political one doesn't mean that any outcome or action even remotely political is inherently the goal of the project. That's an absurd hasty generalization.

Further, just because one of the overall goals of the project is political doesn't mean that every goal of the project is political. That's a composition fallacy.

What arguments don't work?

The truth is that FSF is primarily a political organization. It is more of a political organization than any other open source foundation, eg they do things like refuse to endorse any OS that distributes any piece of non-free (read: GPL) software. They advocate against non-GPL open source as harmful. They guide their projects to be technically worse if it supports the political mission.

There is confusion here about what Stallmans role even is since he isn't an active technical contributor: his role to make sure the political vision is being enforced in the projects under FSF/GNU stewardship and that is exactly what he is doing here.

Basically, if something is not advancing the goals of the FSF then it shouldn't be a part of the FSF. It isn't news that the goal of the FSF isn't to make high quality software and they just think free software is the pragmatically best model to achieve that:

https://www.gnu.org/philosophy/compromise.en.html

https://www.gnu.org/philosophy/when-free-software-isnt-pract...

> The truth is that FSF is primarily a political organization.

So what? That doesn't matter. What part of the FSF[0] or GNU[1] project suggests that abortion is an issue relevant to their mission or goals? Why is abortion a software freedom issue? How about we say RMS does adopt it. Does that benefit the other stated goals of the organizations? How?

What happens when the FSF runs a fund raiser, and they hear, "I'd love to give to your organization because I support free software ideals, but I can't because I disagree with your stance on abortion." Does that benefit free software? What happens when they're working with legislators on bills to re-enforce net neutrality, and the legislator says, "I'd love to help you on net neutrality, but I can't. My constituents are overwhelmingly pro-life, and I've gotten hundreds of voters contacting me since Fox News ran a piece on how your organization is pushing this bill to promote pro-choice laws. The National Right to Life and National Pro-Life Alliance have been calling to lobby against this bill as well. I'm sorry, but it seems that people think net neutrality is a pro-choice issue." Does that benefit free software?

So, is the FSF and GNU willing to sacrifice their software freedom goals in order to further pro-choice goals? Should they be?

0: https://www.fsf.org/about/ 1: https://www.gnu.org/gnu/manifesto.en.html

The issue at hand seems to be about suppression of information (information that happens to be about abortion), not pro-abortion.

> but it's a community project and at some point the consensus will be that he is just wrong.

No, that's just your prejudice talking. The consensus can just as well be that RMS has a point and that a lone maintainer has no legitimacy to dictate policy, let alone override executive decisions. RMS is the founder of many GNU projects and is also the president of the FSF. Just because some dude has access to a repository of a FSF project that doesn't give him the right to hijack the project to suit his fancy. It's a FSF project and the project's driving policy is set by the FSF.

The benefits of the doc is purely subjective (obviously) so asking what purpose it serves is a dick move.

And it specifically says not to remove. RMS only objected when they tried to introduce a change he specifically said not to remove. Calling this an unwarranted exercise in authority is also a dick move.

It isn't even a code change. Let's talk priorities: glibc could see improvements preventing y2038 bugs and fuzzing, but we fret over a joke.

Also there is no evidence that the joke could be traumatic to anyone.

There are a lot of other people who have spent countless hours on glibc, also.