Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!apple!julius.cs.uiuc.edu!wuarchive!zaphod.mps.ohio-state.edu!sdd.hp.com!hplabs!hpfcso!mjs
From: mjs@hpfcso.HP.COM (Marc Sabatella)
Newsgroups: comp.lang.misc
Subject: Re: Re: Aggressive optimization
Message-ID: <8960020@hpfcso.HP.COM>
Date: 30 Oct 90 16:01:16 GMT
References: <2301@wn1.sci.kun.nl>
Organization: Hewlett-Packard, Fort Collins, CO, USA
Lines: 19

>Fancy assemblers have been around for a long time. Consider the
>VAX assembler (the DEC one, not the UNIX one), which had the pseudo-op
>JBR which was translated into a JMP or BRanch instruction depending on
>how far away the target ended up, and allowed a high-level specification
>of the register save mask for subroutine calls, so you didn't have to
>figure it out in binary. Or the MACRO assembler for DEC 20s, which
>had all sorts of high-level stuff. It was practically an HLL.

None of these transformations violate the cardinal rule of assemblers, namely
that of providing a one-to-one translation.  That is, pseudo-ops, registers
masks, and macros are all merely notational conveniences.  If I write a macro
and then use it, I know exactly what code will come out.  Conversely, if I
know I want a specific code sequence to come out, I can write the source
accordingly.  The MIPS "assembler", if the instruction scheduling or whatever
cannot be disabled, fails this test of assembler-ness.

Of course, I have nothing against this, and there is probably no real such
dictionary definition for "assembler".  It is counter-intuitive, though, in a
way that pseudo-ops, masks, and macros are not.