RMS regrets not accepting LLVM into GCC during 2005

https://lists.gnu.org/archive/html/emacs-devel/2015-02/msg00594.html

by jordigh • 11 years ago

82 22 11 years ago

lists.gnu.org

RMS is becoming ever more irrelevant with time.

In the same way BSD software has been ever more present, free software is still going strong, not in the way he likes, though.

fit2rule• 11 years ago

He may be irrelevant, but the cause and the philosophy he espouses is now more relevant - and needed - than ever before.

baldfat• 11 years ago

RMS drives me nuts most of the time (the anti-mono as an example) BUT he is absolutely needed and I am glad he is alive. His ethics and reasoning is needed more now.

crxgames• 11 years ago

I completely agree. I've had the opportunity to meet the man and despite his appearance and unique personality, we need his philosophy on many topics shoved into the forefront of the media fairly often.

pjmlp• 11 years ago

Yeah, Apple, ARM and GPGPU vendors are all happy with LLVM license.

microtonal• 11 years ago

He is becoming more and more relevant. I remember reading this at the turn of the century:

https://www.gnu.org/philosophy/right-to-read.html

And remember thinking yeah right, what a nonsense, no one would accept this. Although it's written slightly overdramatic, here we are, many people are reading DRMed books that cannot be shared. We are renting our DRMed movies and music online. Soon there will be a day where young kids don't even realize you could borrow a movie, book, or album from a friend.

Say about RMS what you want, but he has a tendency to predict problems 10-20 years ahead.

nappy-doo• 11 years ago

Everytime I read a book on my kindle, I think about this essay. In my mind, I look at it like this:

My income now so vastly exceeds the income I had in 2000, that the marginal cost of a book is so small, and not actually having to warehouse the book is worth the cost to me. Additionally, I believe we'll be moving to models more like what Disney has implemented with its movies (where you can buy it once, and watch it anywhere) is what will happen with most/all media.

unsignedint• 11 years ago

It's just scary to see that what I once perceived quite radical coming from him are becoming very valid and real concerns.

Reminded me of this article in 2012: "Richard Stallman Was Right All Along" http://www.osnews.com/story/25469/Richard_Stallman_Was_Right...

handelaar• 11 years ago

It's almost like refusing to be connected to the internet and consequently being an entire day behind important conversations-in-progress could have some sort of downside.

SomeCallMeTim• 11 years ago

He's a day behind, not 10 years. He didn't see the original offer in 2005, and this email was written yesterday.

autoreleasepool2• 11 years ago

It's almost like being so fixated on the "free", to the point that you don't consider the "software", could have negative consequences for free software.

longlivegnu• 11 years ago

oh well

I guess he is being realistic, it does no good to beg and ask. I hope little Richard isn't sad though :(

alecco• 11 years ago

Title implies he knew about the offer.

  > If I had seen it back then, I would not have had the benefit of
  > hindsight, but it would clearly have been a real possibility.

Also, they are talking about a patch to GCC to make it use the LLVM optimizer backend, and not LLVM itself. Reading comprehension, anyone?

https://gcc.gnu.org/ml/gcc/2005-11/msg00888.html

norswap• 11 years ago

Being a day behind does not explain why he missed it though. He should just have becoming aware of it one day later. I'm puzzled.

autoreleasepool2• 11 years ago

He missed the proposal within the context of the current conversation. He is actually responding to an email earlier in the thread that linked the original email from 2005[0].

RMS does't use an Internet browser so he fetches pages and recieves them in email. He received the html file a day after he had already written his response that he self quotes[1].

He wrote the linked email before he replied without actually having viewed the link of the 2005 email. Therefore, he was unaware of its contents.

It makes little sense out of context.

[0] https://lists.gnu.org/archive/html/emacs-devel/2015-02/msg00...

[1] https://lists.gnu.org/archive/html/emacs-devel/2015-02/msg00...

dagss• 11 years ago

The rest of the thread linked to gets into the intentional lack of modularity in GCC. Lots of open source, not-really-for-profit projects are benefits from LLVM in general: Julia, Numba (numerical JIT compiler for Python), Xdress (C++ wrapper generator for Python)...probably many others. Even apart from licensing issues (wanting to stay BSD-compatible), they could not have built on top of GCC.

LLVM offers pluggability and modularity. You can use whatever you want from the compiler in a very flexible way; use it only to parse C headers, or use it only for code generation at run-time.

GCC is the opposite: Modularity and pluggability has been explicitly prevented because GCC maintainers fear that a pluggable and modular GCC would enable people to "take the edge off the GPL".

I guess there is no way around this. For instance, if GCC was to merge the gcc-xml fork which allows you to parse C headers and get the syntax tree as XML back, then somebody could "abuse" that to create a proprietary compiler that uses the GCC parsers.

(OK, there is some modularity in GCC now, as I remember a student project in Cython that plugged into GCC using command line flags injecting a shared library in order to parse C/C++ headers; I'm not sure exactly what the status quo is. But it's certainly a very different world from LLVM.)

phkahler3• 11 years ago

Nope. Richard wanted GCC to remain monolithic to prevent certain commercial activities. LLVM is modular with many reusable components which was unacceptable to RMS. Of course he'd rather have a GPLed modular compiler than a free but non-GPL modular compiler, but I don't think he believed they'd get as far as they did and that's why he didn't accept the offer.

nextos2• 11 years ago

I think creating walled gardens and artificial barriers is always a mistake. This is the perfect example. Intentionally crippling a GNU project is absurd.

phkahler3• 11 years ago

> I think creating walled gardens and artificial barriers is always a mistake. This is the perfect example. Intentionally crippling a GNU project is absurd.

I agree, and I'm sad to see it happened with the flagship software of the GNU project.

tzs• 11 years ago

David Kastrup has a very insightful comment in that discussion, well worth reading [1].

He quotes a speaker he heard at an "IT for startups" talk he found himself at. The speaker talked about the importance of contingency plans for when things go wrong, but also said "the worst mistake a manager can commit is to be unprepared catering for the eventuality of success".

Kastrup then notes:

    So we worked on making GCC a bad choice for
    proprietary vendors.  What do they do?  They manage
    to unite an industry-spanning coalition of different
    parties including academia into creating a free
    software compiler framework.  Free to the degree
    that we could just take it, slap the GPL on it and
    distribute our own version of it (assuming we were
    prepared to dealing with pissing everybody off in
    the process).

    So what's our reaction to our and GCC's role in
    causing a large-scale industry cooperation creating
    a free compiler framework?

    Panic.  We are not prepared to deal with people
    doing on their own terms what we have fought to
    happen for decades.  So now we try to fight those in
    our midst who want to make use of what we have
    fought for.

    And since LLVM is relicensable as GPL if anybody
    bothered to, the only effective means to "fight"
    LLVM is by necessity effective in fighting GPLed
    software in just the same manner.

    It seems sort of pointless to be fighting our spoils
    because they were given to us in the wrong kind of
    ceremony.

[1] https://lists.gnu.org/archive/html/emacs-devel/2015-02/msg00...

RMS regrets not accepting LLVM into GCC during 2005

Related Stories