In the case of KHTML, they never used it in the first place, so it seems like a particularly inappropriate example. I assume you actually meant Webkit? In that case, they spent half a decade and thousands of engineer-years contributing to Webkit, so it doesn't fit the original complaint about not "trying to contribute" either.
November 4, 1998; 26 years ago (KHTML released)
June 7, 2005; 19 years ago (WebKit sourced)
* (C) 1999-2003 Lars Knoll ([email protected])
* (C) 2002-2003 Dirk Mueller ([email protected])
* Copyright (C) 2002, 2006, 2008, 2012 Apple Inc. All rights reserved.
* Copyright (C) 2006 Samuel Weinig ([email protected])
AFAIK Google did not take ownership of gcc, nor did they try to fork it without contributing to the original. They used GCC for a good couple of decades while contributing to it, but eventually switched to a different compiler. The same for clang, they neither "took it over" nor "forked it without trying to contribute".
That's... not even close to what happened?
Historically, LLVM was at one point proposed by Chris Lattner, while he was at Apple, to be upstreamed into GCC (and relicensed to GPL, natch) for use as at the LTO optimization phase, which was declined. For most of its early existence, it used llvm-gcc as the frontend to generate LLVM IR. In the late '00s, serious effort was put into making a new frontend for LLVM IR which we know as clang, primarily by Apple at that point, which become self-hosting in 2009 or 2010. Basically the moment clang becomes self-hosting, everyone jumps ship from using llvm-gcc to using clang to make LLVM IR.
Google shows up around this time, I think primarily motivated by the possibility that Clang offered for mass rewriting capabilities, since it has extraordinarily good location tracking (compared to the other compilers available), which is necessary for good rewriting tools. The other major area of Google's focus at this time is actually MSVC compatibility, and I distinctly remember Chandler talking in one of his presentations that you need to be able to compile code to trust it well enough to rewrite your code, so I think the compatibility story here was mostly (again) for rewriting.
Also around this time, gcc gains proper plugin support, and llvm-gcc is reworked into dragonegg to take advantage of the proper plugin support. But because clang now exists, dragonegg is no longer very interesting, with almost all the residual attempts to use dragonegg essentially being limited to people trying to use it to get LLVM IR out of gfortran, as LLVM had no fully-working Fortran compiler at that point.
Ars is controversial on YC news. Who knew?
I don't see a problem here. Why should google have to deal with the opinions of a maintainer if they can just maintain their own version. Yeah obviously it would be nice if they'd contribute their changes back to the upstream repo but from a business perspective it's often not worth it.
At my company the inverse of this problem happened way more often: We find a problem but the maintainer just doesn't care. For example the backward-cpp library is a good example where the maintainer just isn't that active in the issues. Why wait for him to respond if you can just fork it and keep on moving.
Personally, I take the cathedral/bazaar distinction to indicate different development cadences and philosophies, rather than whether contributions are allowed/encouraged.
Various cathedral-style projects (eg: FreeBSD, Emacs) still actively take contributions and encourage involvement.
There's something even further along the spectrum that's "we provide dumps of source code, but don't really want your patches." I'm not sure what the best term is for that, but "source [merely] available" sometimes has that connotation.
In fact "source available" usually means you can see the source code, but there are severe restrictions on the source, such as no permission to modify the source even for your own use, or no permission to create forks of the project containing the modifications, or severe restrictions on such modifications. An example is MongoDB's Server Side Public License, which is source-available but not open source.
And when they don't when talking about source code, they are wrong. If someone says that an RJ45 cable is "a piece of software" because it's "soft" (you can bend it), would you say it's just a different perspective?
Open source, in the context of software, has a particular meaning. And it is the case that many software developers don't know it, so it's worth teaching them.
I agree that the OSI meaning is worth teaching. But perhaps not by saying "you're wrong; there is only one right way." Perhaps more like "some people attach XYZ specific meaning to that phrase, please be aware of it. Also, here is some history of the term if you like."
----
Aside: On re-reading this, I wonder if it comes across as testy... I think I am just channeling my annoyance with the language police of the world, in general, who sour people's interest in topics with their gatekeeping behavior. I don't mean it too personally towards you (:
> One could ask whether Google works ‘open source’ or more ‘source available’; the source is there but you cannot contribute, if you can build it at all
The author of this comment says "if you can't contribute, shouldn't you consider it `source available` instead of `open source`?".
There is only one valid answer: "No, you should not. It is still open source even if you cannot contribute". The context is clear, we are talking about "open source" vs "source available", which are both very specific in this context.
> I think I am just channeling my annoyance with the language police of the world, in general, who sour people's interest in topics with their gatekeeping behavior. I don't mean it too personally towards you (:
No offense taken, and I don't mean it personally either =). My point is just that in this context, the author of the comment was pretty clearly talking (asking, even?) about the difference between "open source" and "source available".
I don't even think it's shutting down the author: there was no other point than this, so the "thread" started by this author was purely about the meaning of those words.
Inevitably, a laser-minded person talks with a sloshy-bucket person and misunderstandings ensue.
In sloshy-bucket land, I think "open source" has various connotations — a sense of community, encouraged contribution, being able to build it yourself, improve it yourself, etc.
And I think the commenter, in broad strokes, was saying that Google is not upholding those various virtues that are often associated with "open source," so felt the term was not a good (sloshy) fit.
In particular, I do not think they were asking the question you say they were asking.
In this space, it seems like there are both too many terms (so people rather just pick a popular one and over-apply it) and too few (so you can never find one that quite says what you want). Such is life, I guess. Maybe "open sourcey" would be good, to indicate it's talking about a hand-wavy vague "ness" rather than a particular nailed-down definition. "Google isn't being very open sourcey"? ¯\_(ツ)_/¯
Anyway, all this to say: in the ethos of trying to take a charitable interpretation of people's words, I think it's good to consider the bucket-of-paint possibility, before jumping to corrections and yes/no determinations.
----
Edit: It occurs to me that originally I misinterpreted you as being persnickety, when perhaps you were just trying to answer the question you felt they had asked. Sorry!
> And I think the commenter, in broad strokes, was saying that Google is not upholding those various virtues that are often associated with "open source," so felt the term was not a good (sloshy) fit.
Totally valid! And I like the idea of considering the "bucket-of-paint" possibility before saying "no you're wrong". But on the other hand, sometimes it's worth agreeing on the meaning of words while discussing something.
I feel like I actually happen to regularly be on the bucket-of-paint side. I will often simplify the part of the discussion that I feel is not relevant by saying e.g. "okay this solution is bad, so if we look into this other solution we have to think about ...". And sometimes people really care about starting a discussion saying "by saying it's bad, you make it sound like whoever would think about it is stupid, and that's extreme. This solution is not necessarily bad, because in some situations it may work even though it is suboptimal". To which I tend to say "sure, I said it was bad as a way of saying that we seemed to agree that we would focus on the other one".
Until this point it's perfectly fine for me. What frustrates me is when the discussion continues in what I feel sounds like, e.g. "no, I think that your saying it is bad reflects that you disrespect whoever would think about it, and you should never have used that word in the first place. I am not sure I can ever have a meaningful discussion with you now that you used this word in this sentence, even if you later admitted that it was an oversimplification".
Anyway, communication is hard :-)
Here is a concrete reason why Google open source sucks when it comes to contributions and I don't think it can be improved unless Google changes things drastically: (1) an external contributor makes a nice change and a PR on GitHub; (2) the change breaks internal use cases and their tests; (3) the team is unwilling to fix the PR or port the internal test (which may be a test several layers down the dependency tree) to open source.
> making it easy for you to use the projects
Google internally use Blaze, a version of Bazel. It's so ridiculously easy for one team to use another team's project that even just thinking about what the rest of us needs to do to use another project is unloved dreadful work. So people don't make that effort.
I do not see either of these two points changing. Sure there are individuals at Google that really care about open source community, but most don't, and so their project is forever a cathedral not a bazaar.