Concise, opinionated history of the BSD/SystemV split | Hacker News Reader

Concise, opinionated history of the BSD/SystemV split

article.gmane.org

140 pointsakkartik52 comments

Also, inaccurate. Consider just this paragraph:

> Sun customers responded by using alternatives, and the closest thing to a working freely downloadable compiler was an obscure project called "gcc" from the guy who did one of the three main emacs variants (gosmacs from the maintainer of java, xemacs from the maintainer of netscape, and gnu emacs).

The unbundling occurred in 1990. Gcc was hardly obscure. In this USENIX schedule from 1989 ( http://www.informatica.co.cr/unix/research/1989/0612.htm ) you see Stallman gave a tutorial on "Introduction to the Internals of the Gnu C Compiler". It's unlikely they would consider that for obscure software.

Lucid forked emacs, and released lemacs in 1992. (See http://www.xemacs.org/Documentation/21.5/html/internals_3.ht... .) "The initial authors of Lucid Emacs were Matthieu Devin, Harlan Sexton, and Eric Benson, and the work was later taken over by Jamie Zawinski, who became "Mr. Lucid Emacs" for many releases."

Only later was it renamed xemacs. Zawinski writes: "When Lucid went out of business in 1994, and I came to Netscape, I passed the torch for the maintenance of Lucid Emacs to Chuck Thompson (at NCSA) and Ben Wing (at Sun), who renamed it from ``Lucid Emacs'' to ``XEmacs.''" (http://www.jwz.org/doc/lemacs.html ).

Hence, 1) it's not true that gcc was effectively obscure in 1990, 2) there was neither lemacs nor xemacs in 1990, only GNU emacs and Gosling emacs, and 3) xemacs wasn't directly from one of the Netscape developers.

Would you happen to have a pointer to confirm the unbundling happened in 1990 rather than 1987? That seems to be the crux of your disagreement.

Edit: never mind, between bodyfour's comment on this page and http://en.wikipedia.org/wiki/Solaris_%28operating_system%29#..., it looks like Solaris 2 only came out in 1991. Thanks.

It looks like there were two stages to unbundling. The first was to add a new, better, unbundled compiler, and the second was to remove the base one. These seem to have taken place in 1990 and 1992, respective.

Here's a press release dated April 18, 1989: http://simson.net/ref/free_software/compiler_clips.pdf

> Sun Offers New Unbundled C Compiler

> In addition, Sun introduced a new product, Sun C 1.0 -- its first compiler sold separately from SunOS ... A version of the C compiler will continue to be bundled and supported with SunOS but feature enhancements will be made to to the unbundled version only.

Thus, it can't be 1987.

I based the 1990 date because of discussions like http://tech-insider.org/free-software/research/1990/1002.htm...:

> Thank you for raising the issue of the unbundling of the Sun C compiler (SunExpert, June, p. 8).

Based on that 1989 press release, it may be the first unbundled compiler was in 1989, but https://blogs.oracle.com/tatkar/entry/studio_release_names_f... says that 1990 was the year of the "First unbundled rel (from OS)".

Then for the second stage, according to Sun Technical Bulletin July 1992 ftp://ftp.math.ethz.ch/hg/pub/doc/stb/stbjul92.ps :

> Sun does not plan to “bundle” any compiler with the Solaris 2.0 operating system. The C compiler, which has traditionally been included as part of the Sun operating system, will be sold as a separate unbundled product.

That Wikipedia page says Solaris 2 came out on June 1992, not 1991 as you wrote.

Digging around some more, http://www.xemacs.org/Documentation/21.5/html/internals_3.ht... says:

> In 1994, Sun and Lucid agreed to rename Lucid Emacs to XEmacs (a name not favorable to either company); the first release called XEmacs was version 19.11. In June 1994, Lucid folded and Jamie quit to work for the newly formed Mosaic Communications Corp.,

The first version named XEmacs was in September 13, 1994, so still a couple of years after the C compiler was fully unbundled.

My three points remain correct, with the replacement of "1991", "1992", and "1993" in the place of "1990."

It is probably no coincidence that this was around the time ANSI C became a standard.

I've been trying, but I don't see the connection.

At the time they sold several other compilers. "C, C++, FORTRAN and Pascal are priced at U.S. $2,000 quantity one and Modula-2 at U.S. $2,200 quantity one" says http://ftp.lanet.lv/ftp/sun-info/sunflash/1990/Apr/16.13.com... .

C++ wasn't a standard, but it was unbundled. FORTRAN was a standard, and it was also unbundled. How does being ANSI C affect things?

The point being that ANSI C compliance required compiler changes.

How does that affect anything? Unbundling doesn't depend on changing the compiler. I don't see how the business model is affected by ANSI standardization. How would things have been different if C were standardized in 1984 instead?

It may be that vendors uses it as an opportunity to unbundle (eg HP-UX).

I don't understand how it makes a difference. Unbundling is a sales strategy. They could have unbundled even without changing the compiler at all.

In any case, the literature I read (see earlier links) included discussion about improved optimizations, so even without ANSI there was a differentiation between the new unbundled compiler and the old bundled one.

Why do you think it's anything more than a coincidence?

Whether or not gcc was "obscure" in 1990 really depends on which circles you were in, I'm sure, but it's true that it was hardly an unknown project, and that Sun's decision to unbundle the C compiler helped it to become much more popular.

Also missing from the article was a nod to Cygnus Support. A lot of the engineering work for gcc was done by the engineers at Cygnus, who basically collected support fees that were substantially cheaper than the Sun Pro C compiler. To be fair, I doubt Sun at that era had anticipated the rise of the "sell support and custom engineering for free software" business model. Cygnus was one of the first companies who was able to show that it was in fact workable.

http://www.groklaw.net/articlebasic.php?story=20050525231654...

> In 1990 I became Executive Director of the Sun User Group. That December I headed for San Jose for SUG's Eighth Annual Conference and Exhibit. ... The second group was irate because Sun had "unbundled" its software. That is, rather than getting all of Sun's developer tools together, they had to be purchased separately. And of course, they cost more this way.

> But wait. Why purchase the C compiler from Sun, when you could get a better one for less money from the FSF? That's what a large number of Sun's users asked themselves. And the net result was a real jump in CD sales at the FSF. (Several years later, when I organized the Freely Redistributable Software Conference [February 1996] and then was Vice President of the FSF, I realized more fully just how much Sun had benefited the FSF. I'm certain this was not a foreseen consequence.)

I think that something a "large number of ... users" know about cannot meaningfully be called "obscure", else my lack of knowledge about European football/soccer teams makes it an obscure sport.

After his illustrious career driving Sun into the ground Ed Zander proceeded to drive Motorola into the ground. But he's still richer than you.

It's funny because it's true. On all three points!

Some inaccuracies in the timeline. Solaris 2 (the sysv one; the BSD SunOS 4 was after-the-fact renamed to Solaris 1) didn't ship until 1993, not the 80s. Also he says that gcc's support for 68k was important but Solaris never ran on that platform -- Sun made the switch from 68k->SPARC a few years before the BSD->sysv change.

The wider point is correct, but it isn't limited to just Sun. The decision for many UNIX vendors to charge extra for the C compiler was a big reason that gcc caught needed momentum. If that hadn't happened the whole free *nix history might have been very different.

I remember having switched to gcc in 1992 on SunOS (sparc) because of nasty bugs in cc compiler.

Yes, that timeline error stuck out to me as well. Recently I found Solaris 2.1 physical media, and the date on it is 12/92. My goal is to get old GCC running on it, but first I have to get it running at all. My sun4m is a SparcStation 20, so will probably have to find a sun4c or SS10.

Note to anyone trying to image old Sun CD-ROMs: anything below Solaris 2.4 (?) is a CD-ROM with a partition table. If you dd out /dev/cdrom on Linux, you may get only the first partition, in ISO9660 format. You will be missing the small UFS boot partitions. There should be one small UFS partition for each supported arch.

Seems like using /dev/sr0 on Linux works.

> I'd link to the message in google groups, but google crippled their groups search functionality a couple years back so finding the message in their archives even when you know the message id and the exact date is unreasonably difficult these days.

I still don't understand why Google made searching the Usenet archive impossible to do.

Some important bits of computing history lurk in those archives and Google just has it all boxed up. The future people are going to shake their fists and wonder why all that information is really hard to get to.

Search by group to arrive here:

https://groups.google.com/forum/#!forum/comp.os.minix

Search group for "linus posix". On first response, click the 'No really, search for "linus posix" button, to discard the helpful "linux posix" correction.

Look at 23 message subjects lines and arrive here:

https://groups.google.com/d/msg/comp.os.minix/T9SjMGTSpXk/0C...

Perfect? Probably not. Difficult? No.

That convoluted process works for that single exact message.

What if you want to search for the early appearance of some two word phrase on Usenet? This used to be easy. It can't (as far as I can tell) be done now.

And that convoluted process is -considering Google has excellent search engine- a fucking stupid mess.

Sorry about the tone, reading it now it doesn't come across well.

For what it's worth, the same author (Rob Landley) has another opinionated article on the SysV/BSD split which also touches today's OS X and Linux, under the context of init systems: http://landley.net/notes-2014.html#04-09-2014

An interesting read, if bound to ruffle some feathers.

That article clearly omits one of the primary failures of Ubuntu's upstart, which was Canonical's insistence on copyright assignment. This more or less doomed it from being accepted by the other distributions, just as Canonical's control-freak nature over being able to re-license free software under a proprietary license doomed many of their other initiatives.

(Some Ubuntu partisans would claim this was only a perception issue, but the legal framework they insisted on using gave them the right to do it, and given that they were a commercial company, and given that people saw what happened with Sun after it was taken over by Oracle, it was extremely unlikely more than a handful would have ever given Canonical the benefit of the doubt.)

"When Ubuntu failed to convince other distributions to adopt upstart, it launched new questionable initiatives such as the Unity desktop (the classic "microvax" mistake that Microsoft also made with Metro) "

What is the 'microvax' mistake? I'm guessing naming your product too similarly to something else?

> This more or less doomed it from being accepted by the other distributions

Didn't RHEL ship with Upstart?

Yes, RHEL6 + derivatives, so I'm a little confused by that statement.

It's well documented, and even Scott James Remnant, the original author of Upstart, didn't sign the Canonical CLA after leaving Canonical.

https://plus.google.com/+KaySievers/posts/C3chC26khpq

"So [Torvalds] went through the Sun Workstation manuals in the university library and started implementing each system call listed there until he got tired of that and had his system call handler print the call number of the next unhandled call bash tried to make, implemented that, rinse repeat until he had bash working."

This seems culture defining somehow. Linux seems to me to be mix and match and the modern BSDs seem to be more planned. Excellent.

Important point: AT&T owned a substantial portion of SUN when they did the SYSVR4 move.

> AT&T owned a substantial portion of SUN when they did the SYSVR4 move.

The all-caps reminds me: SUN was an acronym at one point; it stood for Stanford University Network.

Funny how he brings up the Bryan Cantrill comment. Many years later, that same guy threatened to fire a guy he didn't even employ for reverting a gender neutral pronoun.

Interesting to see the business side of things at Sun, it's rarely under the spotlight compared to other companies such as Microsoft or Apple.

"... as the project grew beyond a boot sector, he wrapped his assembly as C inlines..."

he = Linus Torvalds project = Linux

To this day, I still think of Joyent's CTO as Bryan "Girl Kisser" Cantrill.

And now I will too :-)

For those who don't know the reference:

https://groups.google.com/group/comp.sys.sun.hardware/msg/24...

it was also linked to in the actual article :-)

If you read Dave Miller (http://en.wikipedia.org/wiki/David_S._Miller)'s post that Cantrill responds to (http://cryptnet.net/mirrors/texts/kissedagirl.html), it's a really well researched critique of Solaris SPARC performance, justifying the existence of a port of the Linux kernel for that platform. Cantrill's job title was 'Solaris Performance', so Sun and Cantrill looked pretty poor by responding with a personal attack.

He's also the guy who's stated that he would fire an employee for using proper English: https://www.joyent.com/blog/the-power-of-a-pronoun

Seems like an utterly charming fellow.

That's actually worse than his "Kiss a girl" gaffe. Hard to believe anyone thinks this person should be in charge of anything. They seem to lack both professionalism and maturity. Plus they jump to all kinds of wild (frankly bonkers) conclusion just as the result of someone thinking it should be "he" and "she" rather than "they" or "it."

> To me, that insistence can only come from one place: that gender—specifically, masculinity—is inextricably linked to software, and that's not an attitude that Joyent tolerates.

... I don't even... I hope he never discovers popular fiction, he might have some kind of mental breakdown with all the "gendered pronouns" (assuming his current state isn't already the result of some kind of mental breakdown).

Why is it hard to believe? He is a technically-skilled person, working on a tech project. Who cares about professionalism and maturity when the code works?

I know Bryan pretty well (we're both ex Sun people). He's not as bad as he can sound (coming from me that's sort of amusing, I've got a bit of a rep as being annoying :)

Bryan is a really smart guy, a lot of depth in systems, both software and hardware. He was also incredibly pro-Sun. I remember spending a day with him in San Francisco, he was still at Sun, I was elsewhere. We had a long discussion about operating systems and hardware that was a lot of fun.

Until we got to Solaris and SPARC. He just refused to hear anything negative about either. I tried to point out that Intel was ahead and he just wouldn't hear it.

I think he let loyalty to his company get ahead of the truth and that was weird.

Not that you aren't frequently a victim of the same, but you're rewriting history a bit here -- I was actually one of the most vocal proponents of x86 within Sun, especially with respect to trying to get the SPARC guys to understand just how deeply non-competitive their microprocessors had become. And to give you my decades-old Rashomon-like account of the day you're referring to, it is my recollection was that I tried to convince you that technologies that we were developing like ZFS and DTrace had technological merit, but you felt so venomous towards Solaris that you simply couldn't hear it. So perhaps we each heard what we thought we wanted to hear. ;)

Hey Bryan, you could be right, I never really got over the Solaris thing. It was such a step backwards (at the time, it had none of the Sun goodness that was in SunOS). So maybe I heard it wrong. But at the time it seemed like you had a blind spot for anything Sun. Which resonated with me on the passion part but not so much on the lets make things better part. Sounds like you got past that.

I can only imagine the fury over the stupid decision to replace SunOS with Solaris: it was corporate wrong-headedness at its absolute worst, and the company was lucky to have survived it at all. I did love Sun[1] (the good parts, anyway) -- and I'm eternally grateful that we managed to get the system open sourced before the ship went under the waves. I still use DTrace, ZFS, zones, etc. every day[2], and that is thanks in no small part to your vision -- had it only been heeded in a more timely fashion!

[1] http://dtrace.org/blogs/bmc/2010/07/25/good-bye-sun/

[2] https://smartos.org/

If I'd not been able to run Sun OS 4.3 was it? for a long time, until Solaris was tolerable, a lot of future Sun purchases influenced by myself would not have happened.

What was most galling was that guinea pig advertisement, crowing about a rather broken current version of Solaris in comparison to the still not yet ruined and much less buggy Windows NT of the time. To those of us with a foot in both ecosystems ... it really rubbed us raw when struggling with Solaris.

Fine, I assume he'll apologize for the treatment of Ben Noordhuis then.

EDIT: And the ministry of truth voting is on the side of some rich silicon valley guys again.

"them/they" is also proper English, and although he and I might differ on how firm we are about the usage of pronouns, I agree with most of what he wrote there and think it's commendable that he's trying to make at least a small difference.

At least one chapter of Beautiful Code that I can recall used the random he/she pronoun strategy, and it's distracting and hard to follow.

Making software development sound less overwhelmingly masculine is probably a good idea.

Can we not forgive and forget a single, isolated mistake made many years ago? I'm sure bcantrill has already been punished quite enough for that one mistake.

The Internet is not especially forgiving. This is a good case study in why it's probably not ideal to quote (especially unattributed) snippets of popular culture[1] to try and convey deeper feelings -- especially frustration -- when you cannot read the room.

[1] http://www.screeninsults.com/saturday-night-live-get-a-life....

Am I the only one who doesn't read it as an insult?

I thought it was a funny way of conceding David Miller's points. A sort of not conceding but actually really conceding and putting on an ironic attempt to save face.

A mimicking of a kids retort when he/she knows they've been beaten in a game or argument.

   Oh yeah?  Yeah?  Well *you* smell!

I should clarify: I'm not hurt or offended by bcantrill's words, and the commercial success of Linux in the intervening years speaks more than any verbal response ever could.

On the other hand, even as a Linux advocate I found bcantrill's post freakin' hilarious. There are very few times when the universe gives you a perfect opportunity for a snappy retort, so I can't blame him for seizing the moment. While it's probably not bcantrill's favourite memory, I think that moment deserves to be remembered and respected for its comedic value, if nothing else.

I don't know... can Bryan stop threatening to fire employees he doesn't employ because they reverted a non-gender neutral pronoun?

Reply on news.ycombinator.com