Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!ucsd!ucbvax!ENG.SUN.COM!Mitch.Bradley
From: Mitch.Bradley@ENG.SUN.COM
Newsgroups: comp.lang.forth
Subject: Re: misc ?s on BASIS15
Message-ID: <9105061917.AA00426@ucbvax.Berkeley.EDU>
Date: 6 May 91 17:55:11 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Reply-To: Mitch.Bradley%ENG.SUN.COM@SCFVM.GSFC.NASA.GOV
Organization: The Internet
Lines: 54

> 8.2.0742 AT-XY
>         Ok, now we can set the cursor, but how do we read it?

You can't.  A lot of terminals don't allow you to read the cursor position,
and to entitle a program to read the cursor position would force the Forth
implementation to interpose a software layer in the output stream that would
model exactly what the terminal is doing.  Hopelessly complicated, and of
questionable value.

> 8.2.1645 GET-DATE
>
>          Where is SET-DATE?

Nowhere.  Presumably the operating system has some means for setting the
correct date, and you only need to do it infrequently.  I have had a SET-DATE
function in my Forth system for many years, and I have NEVER used it, other
than to test that it works.

>         Furthermore I think DATEs should only refer to day/month/year
>         so I suppose GET-TIME (and SET-TIME) for the sec/min/hour would be
>         better. At least it would reduce the ridiculous number of
>         stackentries.

They need to be combined into one function so the application program doesn't
have to go to great lengths to ensure that it is getting a consistent sample.
Suppose you call GET-DMY just before midnight, and the clock ticks before
you call GET-HMS.  The aggregate time that you come up with is wrong by 24
hours.

If the stack entries bother you, you can define whatever function you want
on top of the 6-stack-entry version.

> 15.1.1643 GET-CURRENT
>      1647 GET-ORDER
>      2195 SET-CURRENT
>      2197 SET-ORDER
>
>         Why not use @ and ! prefixes instead of the GET- and SET-?
>         (The same would apply to 8.2.1645)

I put it to a vote at last year's FORML conference, and GET/SET won.  The
@ and ! versions were included in the ballot.

> 16.1.0245 /STRING
>
>           I suppose the defintion should read:
>
>        ...begins n characters beyond c-addr1 and is n characters shorter...
>                  ^                                  ^
>           (In BASIS15 u1 is used at the ^s)

The ability to lengthen a string is fraught with danger.

Mitch.Bradley@Eng.Sun.COM