Dennis Ritchie passed away October 9th. Dennis was the creator of the C language and one of the principle creators of the Unix operating system, and was highly influential in many other aspects of modern computing.
2 plus ones
Shared publicly•View activity
View 9 previous comments
- I remember from somewhere on world about your hello world. I also had a few exchanges thru usenet with him, on the topic of C and BCPL and was quite glad to get responses from him about things. I will miss him.Oct 13, 2011
- The GCOS linkers had a record type (remember those?) that supported 36 bits of BCD characters, for a length of 6. When we integrated the Waterloo C compiler, we used a different record type that allowed us to use a 7-bit ascii string, although I don't remember how long a string we accepted. I think it was 8 (72 bits).Oct 13, 2011
- http://man.cat-v.org/unix-1st/ -- does not show a GECOS field; it does show several sys calls with 6-character names.: I cited 6th Edition because I didn't feel like looking around for something earlier at that hour. That said --
The real question, though, is why GECOS' limitations should have affected Unix system calls' names. Even the original Ritchie and Thompson paper (SOSP, 1973) stresses that the system was self-supporting even from the beginning, and didn't need to be cross-compiled. (Interestingly, that 1st edition manual I cite above has no C compiler, just an assembler, and sections 2 and 3 are written for assembler language programmers.)Oct 13, 2011
- Gosh, I don't know Steve, it would be a good excuse to ask Dennis but...
Anyhow, that's how it came down to me, I was not involved. Maybe it wasn't the GECOS linker but there was a 5 char (plus an external marker, underscore or $) somewhere which was why create got shortened to creat(). Maybe it's not true, or not being fully explained, I don't think reading tea leaves in the 1st ed manual or the 73 paper are going to lead to the answer, we'd have to find someone who was there, most likely.
The real point of that note was the amusing story about the edit session transcript where create() seems to become creat(). Since I was there personally and looked at the transcript because I thought it pretty amazing they'd have something like that I can vouch for what I saw.
Mark Bartelt and Norman Wilson were acting as Masters of Ceremony at that event put on by the "Languages and Tools SIG" which was really the Unix SIG (Special Interest Group) but DECUS did not allow the use of trade names in, well, anything. I got scolded because though I carefully followed those rules in my talk someone stood up at q&a and pointed at my diagram (it was about some piece of our network at Boston Univ which was sort of interesting at the time) and said that box, what is that exactly and I said an Ungermann-Bass blah blah and that was it, I'd mentioned a TRADE NAME! OMFG! So got "talked to" by some DECUS mucky-mucks after the talk, I apologized but thought I could answer a direct question like that and then we all went to dinner (Dennis also) at the Mexican restaurant just across from the Anaheim Convention Center. I don't remember off-hand what I had for dinner but it was probably a combo plate of some sort.Oct 13, 2011
- Did Dennis invent "Hello world" as the canonical example program, or did it predate C or did bwk write it?
And do you know who actually typed 1,$s/create/creat/ on the raffled transcript?Oct 13, 2011
- I have no idea if Dennis invented that example, probably. It was more important than it might seem at first glance because back in the 70s writing that example on many common programming language / OS combos was way more complicated than that. Even where the program itself was relatively simple, such as Fortran (below) it was often pretty inscrutable and on something like OS370 (which accounted for the vast majority of computing environments in the world back then) also required several lines of really inscrutable JCL to compile, link, and run the program.
Since the formatting gets messed up I'll use _ for blanks:
I have no idea who the edit transcript was associated with, but I actually think I don't remember because it must have come up at the time. I'd like to think it's still out there, maybe someone, Mark Bartelt?, knows who ended up with it.Oct 13, 2011
Add a comment...