Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site birtch.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxn!ihnp4!qantel!hplabs!sdcrdcf!trwrb!scgvaxd!felix!birtch!oleg
From: oleg@birtch.UUCP (Oleg Kiselev x258)
Newsgroups: net.unix
Subject: Re: help - escape key detection
Message-ID: <139@birtch.UUCP>
Date: Tue, 1-Oct-85 01:33:13 EDT
Article-I.D.: birtch.139
Posted: Tue Oct  1 01:33:13 1985
Date-Received: Mon, 7-Oct-85 03:41:44 EDT
References: <595@dicomed.UUCP> <65900004@trsvax>
Organization: Birtcher, Santa Ana, Ca.
Lines: 23

>      This is the solution adopted by vi: When an escape is read, the alarm
> clock is set to one second and then a read is issued for the next character.
> If read returns before the alarm goes off, it is assumed that characters
> could have been transmitted that quickly only if they were generated by the
> keyboard as a function key was pressed; if the alarm goes off (interrupting
> read), it is assumed that the characters were typed slowly by a human--that
> is, the escape itself was actually pressed.
> 
> 	[goes on to describe problems with this system; eg. fast typist or
> 	 a slow system could screw this scheme up.]
> ---------------------------------------------------------------------------
> 
> It gives a section of code which would simulate what vi does.
> If you are really interested in this, you should get a copy of this book.
> 					George Moore (gm@trsvax.UUCP)

Oops! Don't lough, but it seems we wasted a few hours writing the same
damned thing... Only we used sititimer(2) ( a 4.2 BSD call - lets you set
intervals in useconds). 

But, seriously, there is a problem with that scheme as well -- some terminals
perform HOME and some other functions on the CRT ONLY -- and don't send
anything to the program... Example - HDS terminals. Or am I missing something?