Happy Birthday, Linus 376
Glyn Moody writes "Today is the birthday of Linus. Just under 19 years ago, on the first day the shops in Helsinki were open after the holidays, Linus rushed out and spent all his Christmas and birthday money on his first PC: a DX33 80386, with 4 Megs of RAM, no co-processor, and a 40 Megabyte hard disc. Today, the kernel he wrote on that system powers 90% of the fastest supercomputers, and is starting to find its way into more and more smartphones — not to mention everything in between. What would the world look like had he spent his money on something else?"
A case of the pundays (Score:5, Informative)
How would the world look different? It would be a whole GNU world.
BTW, Linus is 40 today, there seems to be no mention of that anywhere.
Doesn't qualify for one-name status (Score:4, Informative)
Comment removed (Score:2, Informative)
Re:DX or SX? (Score:4, Informative)
The 386SX was a 32 bit processor internally but had a 16 bit data bus. The 386DX was a straight 32bit processor all the way through. There was a third flawed varient that had a problem switching between real and protected mode that could lock up the system. Those chips would be stamped that they were only certified for 16 bit apps. The ones that tested good had a double sigma stamp on them. Neither the 386sx nor 386dx had math coprocessors. The 486 however was a different story. The 486DX had a coprocessor but the 486SX did not.
No coprocessor... (Score:4, Informative)
Interesting read.
Re:A case of the pundays (Score:5, Informative)
I would counter that, while it inevitably would have happened at some point, it's not a given that the resulting OS would have been GPLed, and subsequently things could've turned out very differently. Happy Birthday, Linus!
Re:No coprocessor... (Score:3, Informative)
The 486 SX was just a 486 that they could not guarantee that the coprocessor would work so it was switched off.
Re:Debian GNU/kFreeBSD (Score:4, Informative)
There's also at least a small chance that many of the kernel hackers who work on Linux today would have been working on the Hurd kernel.
I doubt it would have happened. The Hurd hackers wanted to do fundamental OS research, and everyone else wanted a "Unix" kernel that they could just use and hack around with, and which didn't cost a lot.
I can remember that the biggest factor in our little group of hackers moving to Linux (from 386BSD) was that it had working shared libraries. OK, they sucked in many (many!) ways, but it still meant that you didn't need to have loads of copies of libc in memory or on disk at once. On the small machines of the time, that was a massive saving.
Re:Debian GNU/kFreeBSD (Score:3, Informative)
Because BSD was being disputed in court:
BSD 4.4 was only released in '94, due to the lawsuit, and Linux had already gained popularity.
Re:Doesn't qualify for one-name status (Score:2, Informative)
Linus (Torvalds): Sounds like "Penis"
Re:Debian GNU/kFreeBSD (Score:5, Informative)
Just because of the permissive license?
Partly - commercial BSD derivatives and the BSD networking stack ending up in Windows 95 put off some developers - who wants to see their work being co-opted by Microsoft and other corporations in closed source products? Some people no doubt, but not the majority. Another effect of the licensing was that BSD splintered into different, slightly incompatible commercial forks. The GPL protected Linux from that fate - the free distributions were shipping more-or-less exactly the same kernel code as the commercial distributions - the general perception was that with Linux free didn't mean "less", it just meant "no telephone support". In BSD land, "free" (to many) meant "not as good as what Sun and IBM are selling".
Another factor was that BSD was seen by some Europeans as being controlled by some American labs and American universities. This made it seem less approachable - and harder to get your code into. Linus, in contrast, welcomed good patches with open arms. Linus was highly enthusiastic about people developing code for his kernel. The same enthusiasm for outsiders was not visible within the BSD community.
Yet another factor was the number of Linux distributions that sprung up. Competition is good - Debian, Slackware, Red Hat, etc. were all competing to make the best Linux distribution, and there were numerous other distributions trying to push them from the top. In contrast, BSD was more centrally controlled, and whilst there was some competition between distributions, there wasn't a great amount. Plus the licensing made forks more likely - with GPL and Linux, if someone wrote a good patch, it was highly likely that patch would end up in all Linux distributions fairly rapidly. The same could not be said of BSD and its various commercial forks.
Re:No coprocessor... (Score:3, Informative)
> The 486 SX was just a 486 that they could not guarantee that the coprocessor would work so it was switched off.
Exactly, I thought mostly every /. reader knew that but reading the comments and replies makes me feel a little old... ;-)
In short SX and DX where made at once, then on the testing stage, if the co-processor failed, they sold it as a SX, if it worked well, they sold it as a DX.
Note that this principle is still applied today, I wrote about it previously to explain why overclocking sometimes work and sometimes don't:
http://slashdot.org/comments.pl?sid=1246015&cid=28105441&art_pos=5 [slashdot.org]