GNU sed 4.2.2 released, maintainer resigns

For a time I too was the maintainer for GNU sed. Part of that time I was paid by the FSF for that work (this was a long time ago, when the FSF had a small staff working directly on the Hurd and GNU). When I started, GNU sed was very incompatible with the relatively new Posix standard for sed. When I finished, it was less incompatible. It was during this same time that I started work on a new regexp engine for sed but that was not done by the time I stopped work on sed.

From that position I was able to observe fairly closely how the GNU project was being led, technically, in the days before the Linux kernel had had any real impact.

RMS's technical leadership was, I think, not very skilled. Let me explain what I mean:

If you were working on a program and sought his advice, he was very good at zeroing in on the issues and giving excellent advice. And sometimes if you were working on a program and he noticed something he didn't like about your approach, his criticisms were very good. People used to tell stories about how good a programmer he was and those stories were basically all true. He was sharp and I assume that, in spite of his age, he still is.

The problem was that he showed no effective capacity to really lead the larger meta project of pulling together a complete OS. He tried -- with projects like autoconf and documents like the GNU coding standards. And he kept a list of programs that, once we had those (he reckoned) along with a kernel -- GNU would be "done". That was about the extent of his "big picture" for project management.

Mainly, he concentrated on advocating for the idea of software freedom. I think the gambit was that if enough people demand their freedom, the project of organizing a GNU project would become easier. I don't think this gambit worked.

That was never a clear enough, coherent enough, or informed enough vision of the complete GNU project and, consequently, GNU has never really successfully gelled. You can grab some "100% libre" distributions, these days, but only barely. There is no sustainable culture and technical organization there ("yet", I hope).

The RMS failure I see is a failure at being a community organizer of GNU programmers. A lot of people got the vague idea of a GNU project. Many of us were happily recruited to the goal. But everyone I worked with at the FSF, including me, kind of went off in various incoherent directions -- doing what we guessed would help and that seemed interesting to us. We never "pulled together as a team" and, in the GNU project, that still doesn't happen.

The GNU project gradually accumulated a heck of a lot of very good "parts" but could never gel. The first three world-changing releases (GDB, GCC, and Emacs) really startled people. The various shell/text utilities in those early days spread because they were often usefully a little bit better than the proprietary "native" equivalents shipped by Sun, Dec, AT&T, etc. People sat up and took notice but behind the scenes the project of setting up a lasting "complete OS" project that would promote software freedom for all users ... never quite came together.

The "open source" people -- who I also later worked for, because I made a mistake in trusting them at their personal word to me -- seemed at first like they might help bring resources to the problem. In fact, what they mostly concentrated on was creating proprietary products using the free software "parts" from the incomplete GNU project. In the early days they sought to monopolize some of the key labor for the GNU project (and they succeeded, because they paid much better than RMS and many of those particular hackers didn't really give a shit about the freedom of users). As the "open source" industry matured it perfected its model of a perpetually incomplete / inadequate free software OS as a source of inspiration to enthusiastic youngsters, realized in practioce as a perpetually freedom-denying set of proprietary OS products. Companies like Red Hat and Canonical realized that they could exploit the deficit of community organizing to charge high rents for libre software, so long as they don't care seriously about the freedom of users. That's what they did and what they do.

So in my view, RMS was not good (and still is not good) at leading the GNU project -- but the real tragedy is brought on by the glad-handing, deep-pocketed, "open source" rentiers who place concern for their own profit above the freedom of the community.

proprietary OS products?

> I'm fairly confident the threat of not being able to buy a general purpose computer is not a high probability outcome.

I already have to go very far out of my way and pay a huge premium for getting an unlocked Android phone. As phones/tables supplant many "general-purpose" computer use cases, this threat is only increasing, if current trends continue.

Yes, where he has, for example, refused to allow emacs to use any sane bug tracking system because it would interfere with his workflow, despite the fact that he's not the only developer.

LLVM's developer policy requires any developers contributing patented code to non-assert any patents with that code. This is particularly true if they work for companies. If you know of places where this is not occurring, or people are contributing code that is being aggressively patented, please email me personally and i'll make sure it gets taken care of.

Also note that assignment does not fix anything related to patents. You seem to be confusing contributor agreements in general with copyright assignments.