On Thu, Aug 11, 2011 at 04:00:21PM -0300, Jecel Assumpcao Jr. wrote: > The original Smalltalk-80 from Xerox used CR as its line separation > character, but the really big external influence on Apple was UCSD > Pascal which shared that convention. Apple, however, (along with > Commodore and Tandy/Radio Shack - the big 3 from 1977) had already > adopted this convention from the start.
Fascinating! Thank you! I've been wondering the same thing for years. > DEC had adopted the LF+CR style required by teletype terminals and this > was copied in CP/M, then MS DOS and finally Windows. I know that technically + is commutative, but it's worth pointing out that DEC, Intergalactic Digital Research, and Microsoft all used a CR followed by an LF, not an LF followed by a CR, although either order works on a teletype, or in the terminal emulator I'm using now. > http://en.wikipedia.org/wiki/Newline An excellent article, thank you. It points out that the BBC Micro *did* use an LF followed by a CR. In addition to the various eccentric systems listed on the WP page, ASCII itself has a four- or five-level hierarchy in its control characters (FS, GS, RS, and US, and arguably EOT), and ASCII-1963 (the version in which ^ is uparrow and _ is leftarrow!) had eight or nine. I highly recommend Tom Jennings's excellent history [An annotated history of some character codes or ASCII: American Standard Code for Information Infiltration] [0]. [0]: http://www.wps.com/projects/codes/ The [Pick][] operating system (devoted to business data processing on minicomputers, mostly) is the only thing I know of that used such a thing in its normal file format: a Pick "file" is like a Unix "directory", containing "records" that are like Unix "files". Within each "record" (in 1970s versions of Pick, limited to 32K) "items" are separated by "item marks" (character 255); each item consists of "attributes" separated by "attribute marks" (character 254), and so on with "values" (253) and "subvalues" (252). [Pick]: http://jes.com/pb/pb_wp1.html Getting back to the reinvention of computing: I've never used Pick, although it has a sort of cult following. I've noticed, though, that very simple expedients like this data representation often allow simple things to be done surprisingly simply. You can grep and cat and diff and regexp-extract text files in useful ways surprisingly often. (This was the "Desperate Perl Programmer" desideratum for XML, although I think XML pretty much failed to deliver on that promise.) I think this is what Larry Wall meant by "whipupitude". The tradeoff, of course, is that it's terrifically difficult to write reliable programs that work with Unix text file formats, because it's easy to do "the wrong thing" and have it work by accident on your first thousand tests. Shell scripts' notorious intolerance for filenames with whitespace is probably the best example of this kind of thing. But also, if I grep my access_log, I can grep it without specifying what field I'm interested in --- at the expense of possible false positives. Is it possible to get the whipupitude without the error-proneness, and without sacrificing expressivity? Pick and Unix show the merit of Alan Perlis's epigram, although of course Perlis was not particularly fond of them: > It is better to have 100 functions operate on one data structure than 10 > functions on 10 data structures. Today we might instead say: it is better to have 100 classes that are clients of one protocol than 10 classes as clients of each of 10 protocols. (And that is the reason I am writing this email in vim: because I can use it via ssh inside of screen, and there are thousands of programs that speak the VT100 protocol, all of which can be successfully remoted with ssh and multiplexed and fault-tolerancified with screen.) (I wish I had a neat conclusion to put here. Thoughts?) Kragen _______________________________________________ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
