Is there any evidence of this being true in general? Linux is the only software I can think of where a quality product is accompanied by a "grating attitude".

All of the other software I've liked in recent years (like Redis or SQLite or Postgres) has maintainers who seem rather easy-going.

Those examples don't have the same influx of contributions, and they don't have a host of corporate contributors (Red Hat, Intel, IBM, Google, Nvidia, etc.) who are all vying to push their own solutions into the code. Much of the rage people see from Linus is about these contributions not being good enough because they put the companies' needs before the kernel. The most important quality in a project manager is often the ability to say "no".

You will see Linus rant about what he considers bad code practices, and will castigate people rather brutally for this [1], but he always provides a cogent, often informative explanation. A more mild-mannered lead would not make a scene, and I don't see the point of his expletive-laden gruffness -- it would have been just as effective without the acidity. I guess if you want the Linus who defends code quality from big contributors, you will also have to accept the Linus who verbally destroys individuals.

[1] http://yarchive.net/comp/linux/bool.html

Are there other projects with numerous corporate contributors, then, which were threatened or destroyed by poor patches? My perception is that Postgres has a large number of corporate contributors, but they don't seem to be hurting for it.

This justification (i.e., that Linux has to be this way, because it's in a unique position) sounds strange to me. If we're claiming that the relevant sample size is N=1, isn't that equivalent to saying that it's impossible to study whether this management style is necessary or not?

In fact, this argument seems almost backwards to me. Linux can be run this way because it is successful. Nvidia isn't trying to mess up NetBSD's architecture, precisely because NetBSD is not as popular with users.

> Are there other projects with numerous corporate contributors, then, which were threatened or destroyed by poor patches? My perception is that Postgres has a large number of corporate contributors, but they don't seem to be hurting for it.

There's definitely sometimes strains showing. Corporations have often interest in getting specific features in, but don't care that much about maintaining the codebase and/or maintainability.

Definitely causes some strife. I think we most of the time keep it civil enough.

There are good software maintainers and bad. And good communities and bad. OpenBSD has pretty grating maintainers. FreeNAS has a pretty toxic community. I could go on, but it's certainly not just linux.

In my experience people who give brutally honest and harsh commentary tend to be less offended by it and don't tend to hold grudges the way other people do.

It an open source project where the goal is to promote quality work, I can see this being a valuable attribute.