laser-lovers@uw-beaver (05/29/85)
From: Guy Steele <gls%AQUINAS@THINK.ARPA> I'm sure the ad agencies responsible for the awful typography would think we are merely a small group of irrelevant raving fanatics. Hey, a "g" is a "g", right? But I must note a couple of boners on this mailing list on topics that concern me (despite the fact that the same person happened to send both messages, I'm not out to pick on anyone). First, it was Dan Aykroyd, was it not, rather than Bill Murray [not Murry] who called Jane [not Janet] Curtin an "ignorant slut"? [I am an old SNL fan.] Second, I can hardly wonder at the spread of the flames if a marshmallow-toaster should try to put out a gasoline fire with a bucket of water. [I am a safety nut. I also objected to that TV program a year or two ago that depicted the reaction in an office building to a fire alarm: everyone immediately stopped what they were doing and left the building in a calm and orderly fashion--via the elevators.] Just so this note has some content appropriate to the mailing list: as an author of books about programming languages, I wish that (a) someone would design a really good-looking fixed-width ASCII typeface, and (b) publishers of computer books (particularly my publishers) would use it. I rather like the alphabetic characters used my book "Common Lisp: The Language" (sorry, I don't know the name of the typeface), but the digits stink and the special characters are a hodgepodge amalgamated from several typefaces: the tilde started out life as a swung dash and was raised; " ' < > ^ are much too heavy; # is a bit too heavy and much too slanted; the % is cocked a bit; the | is broken in the center (I prefer an unbroken one) and a bit too heavy; and so on. The ! and ? look good to me. The fixed-width typestyle used in my book (with Sam Harbison) "C: A Reference Manual" has similar but not identical alphabetic characters: the aspect ratio is different (characters of the same height are wider in the C book), and the "a" is a bit different. The digits are much better, and overall the special characters are better matched to the alphabetics, but some are too light, and the braces are too easily confused with parentheses. The Computer Modern \tt font is the best I have seen overall, but some small points annoy me: the "?" is a bit too angular; the "%" has ovals instead of circles (which I prefer); the comma and semicolon should be a bit more distinctly different from the period and colon; the "Q" strikes me as awkward (maybe it's that stiff, straight tail); the tail of the "5" is a bit too closed, making it confusable with a "6"; the top of the "&" is too small, and the tail rises too high. (Points in its favor: a six-pointed asterisk (I dislike five-pointed ones in programs, primarily because APL uses a five-pointed asterisk to denote exponentiation rather than multiplication); "O" and "0" are easily distinguished without resorting to an artificial marker such as a slashed zero; braces, brackets, and parentheses are of similar weight and yet easily distinguished; the "#" is just about perfect.) The CMTT fonts also suffer from severe quantization errors on our QMS printer, causing the N and W in particular to be much too light. The real point I want to make here is that a very different aesthetic is involved in the typographical presentation of programs as opposed to prose. Vertical alignment is much more important, and so either fixed-width typefaces are very careful typesetting of variable-width fonts is required. Characters are used in contexts not at all traditional in ordinary prose, and their looks in such contexts must be considered. Examples: the pairs := -> <= => >= += -= must all look good; one essential property is that the -, +, and = must be centered vertically with respect to each other and the points of the < and >. Consider too the frequently seen LISP idiom `',x where the three "bitty marks" must be clearly distinguishable. And then there is the old problem of striking a compromise on the height (depth) of the underscore, as well as its width: you want it low enough to be able to actually underscore alphabetics, but not so low that it disrupts the baseline in an identifier such as my_hairy_long_PL_I_style_name; and you want it wide enough that several in a row run together, but not so wide that it runs into adjacent non-underscore characters. Does anyone else have any opinions on this program-presentation aesthetic? Please add me directly to the mailing list as "GLS@THINK". --Guy Steele Thinking Machines Corporation
laser-lovers@uw-beaver (05/30/85)
From: trwrb!trwspp!spp3!urban@Berkeley (Mike Urban) As long as someone is criticizing the "amtt" ("cmtt"? are they the same?) font, I'd like to add that the left and right single-quote characters are too hard to distinguish at a glance, and the tilde, which is only rarely used as an accent in this font, is rather difficult to decipher in Unixy examples like " ~uucp/foo ". Probably an alternate tilde character should be available somewhere in this font, or maybe the regular tilde should be lowered and exaggerated, and pray that nobody tries to use it as an accent?
jeff@mcnc.UUCP (Jeffrey Copeland) (05/31/85)
For a typewriter-like font, I am partial to the troff constant width font. (This is basically the font used in Kernighan & Ritchie's "C" book.) It has the full ASCII character set, and it blends nicely with Times Roman, but the only two implementations I know of are for the C/A/T and APS phototypesetters. After that, the TeK tt font is the best, even though it does have weight problems on QMS printers.