Path: ...!weretis.net!feeder9.news.weretis.net!panix!.POSTED.2602:f977:0:1::5!not-for-mail
From: Rich Alderson <news@alderson.users.panix.com>
Newsgroups: alt.folklore.computers,comp.os.linux.misc
Subject: Re: The joy of FORTRAN-like languages
Followup-To: alt.folklore.computers
Date: 01 Oct 2024 16:39:37 -0400
Organization: PANIX Public Access Internet and UNIX, NYC
Lines: 41
Sender: alderson+news@panix5.panix.com
Message-ID: <mddmsjn36d2.fsf@panix5.panix.com>
References: <pan$96411$d204da43$cc34bb91$1fe98651@linux.rocks> <vd8o1s$178gk$5@dont-email.me> <llr46dFmeudU2@mid.individual.net> <vd9r10$1d6gq$4@dont-email.me> <vd9rub$18mq$2@gal.iecc.com> <vd9see$1d6gq$5@dont-email.me> <AF4KO.179529$1m96.177503@fx15.iad> <vdarvh$1l4ch$6@dont-email.me> <mddbk046f1m.fsf@panix5.panix.com> <vdf5fe$2cn51$8@dont-email.me>
Injection-Info: reader1.panix.com; posting-host="2602:f977:0:1::5";
	logging-data="21568"; mail-complaints-to="abuse@panix.com"
X-Newsreader: Gnus v5.7/Emacs 22.3
Bytes: 2927

Lawrence D'Oliveiro <ldo@nz.invalid> writes:

> On 30 Sep 2024 16:51:33 -0400, Rich Alderson wrote:

>> [PDP-6/PDP-10:] Lovely machines to program.

> Bit before my time (the main campus machine was a PDP-11/70 when I started 
> first year Comp Sci). But I know they had a very loyal community, who were 
> very upset when DEC cancelled its "Jupiter" project in favour of going 
> all-in on VAX.

> I read some docs once. One thing I can remember is the "JUMP" instruction 
> is actually a no-op: if you want an unconditional jump, for example, you 
> use "JUMPA", for "jump always". ;)

The instruction set is highly fleshed out, so that there are a number of no-ops.
The usual branch instruction is JRST "jump and restore PC flags"; JUMPA is most
often used by DDT for entering and returning from debugger inserted patches,
although a programmer might want to use it for reasons.

The JUMP no-op is actually useful in TOPS-20, where it is used for various
error catches after system calls.  The action taken depends on the AC field of
the instruction.  NB:  These are not mandatory; the programmer can write her
own code to deal with error returns from system calls, and that is frequently
what is done.

Early in the history of the family, different no-ops were faster or slower, so
choice was important.  In the KL-10 and all following processors (including
those from other manufacturers), the microcode for most no-ops is the same so
does not matter[1].

The one I use most often is TRN "test right half of AC with no bits modified
and do not skip", one of the test-and-skip family.

[1] See above for JUMP with nonzero AC field.

-- 
Rich Alderson					  news@alderson.users.panix.com
      Audendum est, et veritas investiganda; quam etiamsi non assequamur,
	  omnino tamen proprius, quam nunc sumus, ad eam perveniemus.
									--Galen