Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1a 7/7/83; site rlgvax.UUCP Path: utzoo!linus!decvax!harpo!seismo!rlgvax!guy From: guy@rlgvax.UUCP (Guy Harris) Newsgroups: net.cog-eng Subject: Re: Re: thousands of input flags in one - (nf) Message-ID: <1081@rlgvax.UUCP> Date: Wed, 31-Aug-83 19:25:13 EDT Article-I.D.: rlgvax.1081 Posted: Wed Aug 31 19:25:13 1983 Date-Received: Thu, 1-Sep-83 20:26:35 EDT References: <1306@whuxlb.UUCP> Organization: CCI Office Systems Group, Reston, VA Lines: 34 Perhaps UNIX systems should provide a variety of user interfaces, as they definitely aren't just for programmers anymore. However, for all the the flak UNIX gets for its user interface it's not clear that most other *comparable* systems are better by all that much. Would you say that CP/M's, or RSX-11's, or TOPS-10's "PIP" is less "cryptic" than UNIX's "cp"? (Yes, I know that DEC's command languages have picked "COPY" for the copy command now.) Perhaps UNIX's command language isn't the nicest in the world, but are most of the comparative systems *that* much better? If not, perhaps the criticisms should be levelled equally at all such systems. Then again, there was some research at Bell Labs that indicated that the difference in "user-friendliness" between "cp" and "copy" didn't amount to much, if anything - users could pick up whatever the "local" name for a command was in a short period of time. I suspect what's referred to as the "principle of least surprise" is more important; keep your command language reasonably self-consistent (e.g., if the "-v" flag means "verbose" in most commands try to make it mean that in as many commands as possible) and consistent with some fairly straightforward model of the world (i.e., a "copy" command which took the input file as "/INPUTFILE=xxx" and prompted you for the output file might be poor). Complete and *obligatory* terseness is *not* a virtue. The pre-UNIX/TS "ed" which had *no way* of explaining an error message was not the right way to do things, 300 baud terminal or no. Many editors which were used at 300 baud had more helpful error messages. One idea which TECO had and which appears in the current "ed" is that the editor can either output long error messages or it can output short ones and you can ask for an explanation of the last error. If you have to look at the past 20 lines of output to figure out why something went wrong (i.e., because a program is *always* silent about missing files) that's not a good idea. (And "pr"s "Very funny." error message is amusing to those who know what's so funny but insulting to those who don't.) Guy Harris {seismo,mcnc,we13,brl-bmd,allegra}!rlgvax!guy