for the tl,dr crowd:<p>It was noticed a couple parsers included in the source tarball are generated from grammar files that are not. This means they are distributing the compiled version without the corresponding sources (although the parsers <i>are</i> human readable) and that is unacceptable. The situation is analogous to distributing a compiled program without its source or, say, a merged-minified version of a JavaScript library without its original corresponding files.<p>The solution will be applied retroactively to source tarballs and previous tarballs that did not distribute these source files will be removed.<p>It's like the Emacs team just noticed the have been driving their car with a broken headlight for the past couple miles.<p>In the meantime, I seriously doubt many other teams would be so principled as that. Kudos for them.
Cue GOB's "I've made a huge mistake."<p>As much as RMS gets on my nerves sometimes, I have to admire him here. I feel like his reactions are over the top sometimes, but this shows me that at least they're consistently so -- even when he's the one at fault.
In fairness, this is stretching the meaning of `binary blobs' a little, right?<p>I may be misreading, but the issue is that some of the elisp sources of CEDET were generated from bison input files automatically at some point -- but they're still elisp sources; a user can still modify them if needed, and if an asteroid hits the house of whoever did the original conversion, development can still continue that way.<p>That doesn't mean that _technically_ we wouldn't be a lot better off if the bison sources made it into the next emacs release, for completeness and ease of future development -- but I'm not sure I see the GPL violation here.
I thought that you didn't have to distribute the source with the binary so long as you made the source available to anyone that go the binary from you. I.e. if I distribute a binary to someone and they email me looking for the source, I have to give it to them (also I think that you have to make it clear in the distribution that this is the case, though just mentioning that the software is GPL is probably enough).
"We have made a very bad mistake."<p>While I agree with this conclusion and that it needs to be fixed: Wouldn't the irony in the mistake have been a good opportunity to show some sense of humor?
OK, it seems like it may have been an oversight (as in, not to hide anything or subvert someone else). The (over-)reaction by RMS may be a bit too harsh, when a simple README or something might suffice.<p>Every church has to have its pope, though ...<p>Edit: a simple README to point the issue in older tarballs might suffice -- though the issue has to be fixed going forward, no question on that.
> Anyone redistributing those versions is violating the GPL, through no fault of his own.<p>This is not correct and I would expect RMS to call him out on it. You <i>can</i> distribute binaries without the sources. You just have to make them available when someone asks for them in writing. It seem that this is the first time that someone has asked for them, and they are being provided.