Title adapted from the prior submission [1] ("RMS regrets not accepting LLVM into GCC during 2005") but reworded so that it's clear RMS didn't know its contents beforehand.
(EDIT: My original title was "RMS regrets missing an email for the 2005 LLVM/GCC Integration Proposal (2015)", which was reworded by the moderator to match the earlier submission. I don't exactly like the rewording, but am fine with that.)
The actual Integration Proposal [2] proposed an adapter from GIMPLE to LLVM BC so that GCC can make use of LLVM optimization passes without much hassle. Lattner (on behalf of Apple) was willing to assign the copyright of the entire LLVM once it became a preferred backend of GCC. Remember this was before GPLv3 (2007). It is not hard to imagine an alternative universe where this actually happened and GCC remains the dominant free/libre compiler.
I understood Apple choose and funded LLVM explicitly because they wanted to decrease their dependency on GPL-licensed software. If LLVM moved under the FSF-flag, that funding would presumably dry up, or Apple would fork LLVM. I don't see the alternative universe happening.
Besides, GCC fell a bit asleep once until EGCS kicked them awake, and again until LLVM appeared. A bit of competition seems healthy to keep everyone awake, and having both projects managed by the same organisation is a risk here. To be clear, GCC is and stays a great product and LLVM probably needs GCC as much as the other way around.
> > "If people are seriously in favor of LLVM being a long-term part of GCC, I personally believe that the LLVM community would agree to assign the copyright of LLVM itself to the FSF and we can work through these details."
> I am stunned to see that we had this offer.
I'm also stunned to see this offer, in a not-so-delighted way. "If people are seriously in favor of LLVM being a long-term part of GCC" – of course that was a big if, but imagine it happened. Would all the languages that flourished around LLVM (e.g. HN darling Rust) happen in the alternative copylefted LLVM universe?
He says "when I saw your message. At that time all I could see was what was in your message. At the same time I sent another message to get a copy of https://gcc.gnu.org/ml/gcc/2005-11/msg00888.html. That arrived subsequently."
Doesn't that mean he received the original posting after all?
> Oh well.
If you ask me that shows poor leadership. Everybody does make mistakes but not reflecting on those mistakes and trying to do better will result in getting left behind.
> Doesn't that mean he received the original posting after all?
I assume that he received the original posting but wasn't aware of the contents ("I am stunned to see that we had this offer", note the use of tenses). FWIW he uses an email to fetch emails from the external inbox, so it seems likely that it just got shelved for a decade.
I think it's fair. GCC made their decision many years ago. Even this email is 6 years old. Why post it now? Methinks because it fits a narrative that RMS is an ineffective leader that wouldn't know a good decision if it landed on his face.
The original poster here. There is a popular belief that GCC made a strategic decision to make it harder to embed to thwart proprietary softwares and this is all RMS's fault. This posting shows that it's partly false, as RMS would like to accept the proposal if he had seen the proposal earlier (I assume he received but wasn't aware of the actual proposal). I was unaware of this posting before, was amused, checked that there are no recent resubmissions, and submitted. I don't expect you to believe me though.
I thought it was more connected with the recent spree of discussion about replacing C (and GCC) with Rust (and LLVM) in GNU/Linux. A few weeks back this semed to start with a dependency on a Rust compiler by the Python cryptography module, which most notably infuriated the Gentoo community because it meant things broke on some niche architectures not supported by LLVM. There's been a number of articles since then about replacing core utils with Rust implementations and just yesterday an interview with Linus Torvalds about the future potential for using Rust and LLVM for the Linux kernel.
RMS own words and actions have done far more harm to any "good reputation" he might have had, way before anyone on HN... posted links to those words and actions.
(EDIT: My original title was "RMS regrets missing an email for the 2005 LLVM/GCC Integration Proposal (2015)", which was reworded by the moderator to match the earlier submission. I don't exactly like the rewording, but am fine with that.)
The actual Integration Proposal [2] proposed an adapter from GIMPLE to LLVM BC so that GCC can make use of LLVM optimization passes without much hassle. Lattner (on behalf of Apple) was willing to assign the copyright of the entire LLVM once it became a preferred backend of GCC. Remember this was before GPLv3 (2007). It is not hard to imagine an alternative universe where this actually happened and GCC remains the dominant free/libre compiler.
[1] https://news.ycombinator.com/item?id=9028738
[2] https://gcc.gnu.org/legacy-ml/gcc/2005-11/msg00888.html