words:standard_code_that_lets_you_postpone_multiple_commands_without_typing_postpone
Standard code that lets you POSTPONE multiple commands without typing POSTPONE in front of each of them
From willett!dwp@vax.cs.pitt.edu Wed Jan 26 02:42:39 1994 Return-Path: <willett!dwp@vax.cs.pitt.edu> Received: from vax.cs.pitt.edu by ks.mpi-dortmund.mpg.de (4.1/SMI-4.1MHS-mpi-1.4.93) id AA13400; Wed, 26 Jan 94 02:42:30 +0100 Received: from willett.UUCP by vax.cs.pitt.edu (5.65/1.14) id AA16582; Tue, 25 Jan 94 20:30:36 -0500 Date: Tue, 25 Jan 94 06:53:38 EDT From: dwp@willett.pgh.pa.us (Doug Philips) Subject: Part 1 of 1 of POSTPONS.TXT Message-Id: <9401250653.0.UUL1.3#5129@willett.pgh.pa.us> To: plewe@mpi-dortmund.mpg.de Content-Length: 3439 X-Lines: 90 Status: RO } Here is Standard code that lets you POSTPONE multiple commands without typing POSTPONE in front of each of them. { : POSTPONE# POSTPONE LITERAL ; : ]] BEGIN ( ) >IN @ BL WORD DUP C@ WHILE ( >in ca ) DUP COUNT S" [[" COMPARE WHILE ( >in ca ) FIND IF ( >in xt ) DROP >IN ! POSTPONE POSTPONE ELSE ( >in ca ) 0 0 ROT COUNT >NUMBER ABORT" can't ]]" ( >in d ca) 2DROP POSTPONE# POSTPONE POSTPONE# DROP THEN REPEAT ( >in ca ) THEN 2DROP ; IMMEDIATE } This has the following environmental dependencies: It will not handle double or floating point numbers. COMPARE is required by the String wordset, and will not be present in every system. In some systems FIND may not find compile-only commands, and in some systems it may find interpreted commands with the same names which have different actions. In the one case you get an error message, while in the other you get a bug. So this is only 100% portable among systems in which, while compiling, FIND finds the execution token for the compilation behavior -- the same behavior that POSTPONE postpones. Some minimal systems may make the block buffer invalid when you do any parsing operation. ( LOAD will take care of that for itself, but when _you_ do it you're on your own.) ]] does WORD which parses. So ]] is not portable to those minimal systems. This is used in the following form: : LOOP ]] 1 +LOOP [[ ; IMMEDIATE When you do LOOP inside a definition, the 1 gets compiled as a literal and +LOOP gets executed. You can use this for macros -- lists of commands that all get postponed -- without having to type POSTPONE in front of each of them. The code is as long as if you did type POSTPONE in front of each of them, though, and POSTPONE LITERAL after each number. I use such macros to make it easier to type sophisticated control structures. The usual methods don't work to debug such things, so I try to keep them simple -- if there's any error they lose me time. Here is a simple example: : ?EXIT ( f -- ) ]] IF EXIT THEN [[ ; IMMEDIATE You can't portably do R> DROP in a Standard program. There's no guarantee that popping the return stack will have the result of skipping out of the next nested command. This is a limitation on programs which allows much greater flexibility in systems. Well, but how can a command slip out of the command that called it? It can return a flag, and the outer command does EXIT. It's clumsy, but it works. And ?EXIT keeps me from having to write IF EXIT THEN each time, I can just type ?EXIT . And ]] keeps me from having to type POSTPONE three times. I don't know whether ]] is worth using. Maybe the code it produces is harder to debug and harder to maintain. But if it is worth using, almost every standard compiler that has COMPARE can handle it. It's portable. : POSTPONE# POSTPONE LITERAL ; : [[ ABORT" [[ used without ]] " ; IMMEDIATE : ]] BEGIN ( ) >IN @ BL WORD DUP C@ WHILE ( >in ca ) FIND IF ( >in xt ) ['] [[ = IF DROP EXIT ELSE ( >in ) >IN ! POSTPONE POSTPONE THEN ELSE ( >in ca ) 0 0 ROT COUNT >NUMBER ABORT" can't ]]" ( >in d ca) DROP POSTPONE# POSTPONE POSTPONE# DROP THEN REPEAT ( >in ca ) 2DROP ; IMMEDIATE Come to think of it, something like this code should work, without the COMPARE . Someday I'll debug it.
words/standard_code_that_lets_you_postpone_multiple_commands_without_typing_postpone.txt · Zuletzt geändert: 2011-05-10 16:10 von 127.0.0.1