Hall: IBM, Red Hat and Free Software: An old maddog’s view

Hall: IBM, Red Hat and Free Software: An old maddog’s view

> Remind me how it worked — did they provide software under a different license to paying customers and later GPLv2 the same codebase?

They couldn't do that. RMS write GCC initially and Cygnus had no rights to relicense it.

But their paying customers were hardware manufacturers who paid for the compiler for their CPUs. Cygnus would provide them with sources and binaries but it only included sources for these CPUs and not the whole thing.

Free GCC was receiving all the patches but it was never available in binary form for most exotic chips, you had to get these from hardware maker. And it took a loooooong time for some patches to arrive in trunk.

And most of the time these were only coming in binary form, but that wasn't Cygnus problem.

And it took a loooooong time for some patches to arrive in GCC's trunk, so you couldn't pull CentOS on them and just build GCC from sources.

The closest analogue to what RedHat is doing happened with CygWin: CygWin betas are freely available while releases are only available on CD. Reminds you something?

Of course the fact that CygWin only ever released one proprietary version on CD, version 1.0, weakens the point, but that's just because there were too few buyers, even after it become apparent that most users are happy with free betas CygWin 1.0 ISO haven't arrived on any popular web sites and there were no clones.

And the most important fact is that it all was critical for success of GNU: it's unclear if we would even got i386 compiler without all that activity. And without i386 GCC compiler there would have been no Linux, obviously.