🚀 go-pugleaf

>Don't forget the acronym TLA -- Three Letter Acronym.
>DEC is the master (or a master) of TLA's, what with VAX, RSX, VMS, DCL,...

Yeahhh.  I even once saw a fun DEC pamphlet listing them, but I can't
remember the name.

In article <1990Mar26....@eng.umd.edu>, al...@eng.umd.edu (Allon Stern) writes:
> Don't forget the acronym TLA -- Three Letter Acronym.
> DEC is the master (or a master) of TLA's, what with VAX, RSX, VMS, DCL,...
> 
>                              - -= Allon =- -
> 

Just take a look at the assembler mnemonics for the 6502 microprocessor;
   If I'm not mistaken, every single one of these buggers is 3 characters
   (and probably the 8008, 8080, etc. too).

My WAG is that close to 98% of all op-codes available in assemblers that
generate machine code are 3 characters.  The longer/shorter op-codes
only seem to appear in more recent micro-processor development, starting
with the Intel 8088/8086 ( I think, OK its not that recent, ;^) but not
as old as the 6502 etal ).

Wasn't the reason for using a fixed length of 3 for op-codes to make
assemblers easier to implement on computers with very limited memory.
It made parsing commands easier, while being minimaly descriptive.
They could have used 2 character mnemonics, but they would have been
too cryptic.  They could have used 4 character mnemonics, but that
would increase the program source size (not the executable), yet the
op-codes would still be only minimaly descriptive.

Was DEC's use of TLA's influenced by the preponderance of 3 character op-codes,
or did people create 3 character op-codes because DEC used TLA's.
-- 
------------------------------------------------+---------------------
Maybe I shouldn't have done it, sarcasm is so   | Joseph G. Toth Jr.
seldom understood.  Don't FLAME on me, please.  | uunet!tellab5!toth

In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:

>   My WAG is that close to 98% of all op-codes available in
>   assemblers that generate machine code are 3 characters.
>   The longer/shorter op-codes only seem to appear in more
>   recent micro-processor development, starting with the
>   Intel 8088/8086 ( I think, OK its not that recent, ;^)
>   but not as old as the 6502 etal ).

Oh my.  IBM 370 assembly language has mnemonics with lengths
between 1 (N, A, et al.) and 5 (STCTL et al.) and has had
such for a very long time indeed.  You microcomputer people
are all the same. 
--

    "Oh, what a great genius this Pococurante is!  Nothing can please him."
Mark-Jason Dominus 	   ent...@pawl.rpi.EDU	     uunet!inco!alembic!entropy

In article <ENTROPY.90...@pawl22.pawl.rpi.edu> ent...@pawl22.pawl.rpi.edu (Angel of Dope) writes:
|In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:
|>   My WAG is that close to 98% of all op-codes available in
|>   assemblers that generate machine code are 3 characters.
|Oh my.  IBM 370 assembly language has mnemonics with lengths
|between 1 (N, A, et al.) and 5 (STCTL et al.) and has had
|such for a very long time indeed.  You microcomputer people
|are all the same. 

Oh, right.  370 indeed (not even 360!).

Go back to 1401 Autocoder.  Or PDP-1 assembler.  I expect that very
few assemblers had strictly three character op-codes, but some of the
really old ones may have.
=
Nets: le...@bbn.com  |  "There were sweetheart roses on Yancey Wilmerding's
 or {...}!bbn!levin  |  bureau that morning.  Wide-eyed and distraught, she
POTS: (617)873-3463  |  stood with all her faculties rooted to the floor."

In article <54...@bbn.COM> le...@BBN.COM (Joel B Levin) writes:
|In article <ENTROPY.90...@pawl22.pawl.rpi.edu> ent...@pawl22.pawl.rpi.edu (Angel of Dope) writes:
||In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:
||>   My WAG is that close to 98% of all op-codes available in
||>   assemblers that generate machine code are 3 characters.
||Oh my.  IBM 370 assembly language has mnemonics with lengths
||between 1 (N, A, et al.) and 5 (STCTL et al.) and has had
||such for a very long time indeed.  You microcomputer people
||are all the same. 
|
|Oh, right.  370 indeed (not even 360!).
|
|Go back to 1401 Autocoder.  Or PDP-1 assembler.  I expect that very
|few assemblers had strictly three character op-codes, but some of the
|really old ones may have.

Time out!  The opcode is a number, the letters are a mnemonic device so
the programmer can avoid remembering the number.  There is no necessary
correlation between the number of letters in the mnemonic and the number
it represents.

In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:
>In article <1990Mar26....@eng.umd.edu>, al...@eng.umd.edu (Allon Stern) writes:
>> Don't forget the acronym TLA -- Three Letter Acronym.
>> DEC is the master (or a master) of TLA's, what with VAX, RSX, VMS, DCL,...
>> 
>>                              - -= Allon =- -

DEC is A master, I'm sure.  HP is another.  I'm sure most every company
has its own TLA's.

>Was DEC's use of TLA's influenced by the preponderance of 3 character op-codes,
>or did people create 3 character op-codes because DEC used TLA's.

Or maybe HP and DEC and others' TLA's have nothing to opcodes.

HP (or at least the site I worked at) has a wonderful booklet that they
give to new employees which explains all of the TLA's that abound at HP.
It is broken up by products (computers, instruments, etc) and
departments (finance, r&d, personnel, etc).  I remember it being about
15 pages long on 8.5 x 11 paper.  It came in handy when reading the HP
internal netnews.

Matt Squires
squ...@eecs.nwu.edu

In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:
>Just take a look at the assembler mnemonics for the 6502 microprocessor;
>   If I'm not mistaken, every single one of these buggers is 3 characters
>   (and probably the 8008, 8080, etc. too).

Nope.  The 8080 had LHLD (Load HL Direct, I believe).  Haven't the damnedest
idea what it did anymore but I do remember using it.

							spl (the p stands for
							process *this* sucka!)

--
Steve Lamont, sciViGuy	(919) 248-1120		EMail:	s...@ncsc.org
NCSC, Box 12732, Research Triangle Park, NC 27709
"...though you may have the falcon yet we certainly have you."
				Dashiell Hammett, _The Maltese Falcon_

I'd have to say the king of TLA's is probably Control Data.  Their
assembler for their 6600 and greater machines had sort of TLA's:
the first char was operation the second was a register type and the
third was register number (e.g.  BX1    X2+X3   (add x2 to x3 result
in X1)).  Worse than that the archetecture of the machine (big
sucker main processor with a bunch of PDP-8 class machines called
peripheral processors) had the function call to a peripheral processor
be to load the name of the program in its input register then the PP
would go to disk load the requested program and execute it.  Unfortunantly
the program name was 18 bits long (3 6 bit characters).  So all the
PP function codes in the operating system were 3 char long (always 
acronyms for 3 words which seemed to be in the form adjective adjective 
noun)  Examples:  PFM (Permanent File Manager),  CPM (Control Point 
Manager)...  

And since this scheme worked so well for the guts of the OS all 
the rest of the system code seemed to be named similarly.

In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:
>In article <1990Mar26....@eng.umd.edu>, al...@eng.umd.edu (Allon Stern) writes:
>> Don't forget the acronym TLA -- Three Letter Acronym.
>> DEC is the master (or a master) of TLA's, what with VAX, RSX, VMS, DCL,...
>> 
>>                              - -= Allon =- -
>> 
>
>Just take a look at the assembler mnemonics for the 6502 microprocessor;
>   If I'm not mistaken, every single one of these buggers is 3 characters
>   (and probably the 8008, 8080, etc. too).
>
>My WAG is that close to 98% of all op-codes available in assemblers that
>generate machine code are 3 characters.  The longer/shorter op-codes
>only seem to appear in more recent micro-processor development, starting
>with the Intel 8088/8086 ( I think, OK its not that recent, ;^) but not
>as old as the 6502 etal ).

Well, the Z-80 did not have it too much fixed at 3 characters, albeit many
were 3 characters and also matched the same op-codes of other processors
at the time.  (But two of the most used op-codes, LD (load), and PUSH
(just guess that one) )

/----------------------------------------------------------------------\
|      /// Michael Sinz -- CATS/Amiga Software Engineer                |
|     ///  PHONE 215-431-9422  UUCP ( uunet | rutgers ) !cbmvax!mks    |
|    ///                                                               |
|\\\///          "I don't think so," said Ren'e Descartes.             |
| \XX/                    Just then, he vanished.                      |
\----------------------------------------------------------------------/

In article <54...@bu.edu.bu.edu> a...@cs.bu.edu (Al Thompson) writes:
>In article <54...@bbn.COM> le...@BBN.COM (Joel B Levin) writes:
>|In article <ENTROPY.90...@pawl22.pawl.rpi.edu> ent...@pawl22.pawl.rpi.edu (Angel of Dope) writes:
>||In article <23...@tellab5.tellabs.com> to...@tellab5.tellabs.com (Joseph G. Toth Jr.) writes:
>||>   My WAG is that close to 98% of all op-codes available in
>||>   assemblers that generate machine code are 3 characters.
>||Oh my.  IBM 370 assembly language has mnemonics with lengths
>||between 1 (N, A, et al.) and 5 (STCTL et al.) and has had
>||such for a very long time indeed.  You microcomputer people
>||are all the same. 
>|
>|Oh, right.  370 indeed (not even 360!).
>|
>|Go back to 1401 Autocoder.  Or PDP-1 assembler.  I expect that very
>|few assemblers had strictly three character op-codes, but some of the
>|really old ones may have.
>
>Time out!  The opcode is a number, the letters are a mnemonic device so
>the programmer can avoid remembering the number.  There is no necessary
>correlation between the number of letters in the mnemonic and the number
>it represents.

Yes, however, due to memory limits, many of the assemblers were written
such that 3-bytes for the op-code with the 4th byte containing some data
on how to build that instruction.  It made the search/parse code really
simple.  (My PDP-11 assembler had that until Macro-11 came out and was
just tons better and needed a disk drive to work at reasonable rates)

/----------------------------------------------------------------------\
|      /// Michael Sinz -- CATS/Amiga Software Engineer                |
|     ///  PHONE 215-431-9422  UUCP ( uunet | rutgers ) !cbmvax!mks    |
|    ///                                                               |
|\\\///          "I don't think so," said Ren'e Descartes.             |
| \XX/                    Just then, he vanished.                      |
\----------------------------------------------------------------------/

In article <10...@cbmvax.commodore.com> mks@cbmvax (Michael Sinz - CATS) writes:
|In article <54...@bu.edu.bu.edu> a...@cs.bu.edu (Al Thompson) writes:
[...]
|>
|>Time out!  The opcode is a number, the letters are a mnemonic device so
|>the programmer can avoid remembering the number.  There is no necessary
|>correlation between the number of letters in the mnemonic and the number
|>it represents.
|
|Yes, however, due to memory limits, many of the assemblers were written
|such that 3-bytes for the op-code with the 4th byte containing some data
|on how to build that instruction.  It made the search/parse code really
|simple.  (My PDP-11 assembler had that until Macro-11 came out and was
|just tons better and needed a disk drive to work at reasonable rates)

Sure, but that is an engineering decision.  But, there is no necessary
correlation between the size of the mnemonic and the opcode.

In article <54...@bbn.COM> le...@BBN.COM (Joel B Levin) writes:
> I expect that very
>few assemblers had strictly three character op-codes, but some of the
>really old ones may have.

6502 assembler mnemonics were all 3 letters long, although the 
load instructions (and a couple of others) used different mnemonics 
to indicate the register to use (eg. LDA foo instead of LD A,foo).
I suspect the reason for this was that for the load instruction, say,
the addressing modes that were allowed depended on the register
being loaded.  (And the different mnemonics (and different
opcodes, too) probably made writing assemblers easier.)

Scott Lindhurst
scli...@phoenix.Princeton.EDU

I suspect that assembly languages have three letter mnemonics is because HAL
(there's another one) in 2001 used them all the time.  But maybe it's the other
way around...

Maybe they're a hold over from the Depression: WPA, TVA, etc.  Or is it the
entertainment industry invading the computer industry: CBS, NBC, ABC, MGM,
FOX.  The post office (ZIP)?

Whatever.  I have to get cash from the ATM.

Randall Cook

rc...@eagle.wesleyan.edu

	We obviously need a larger alphabet, since there are nowhere near
enough three letter acronyms to go around.  After all, 26^3 is only 17576,
and that's not even enough to give a TLA to each VAX machine instruction,
let alone cover all of the commercial and military applications.  Maybe we
should follow U*ix's lead and make acronyms case sensitive.  That will buy
us a factor of eight in the number of distinct TLAs.

	This is a serious problem.  Why, only last week, I spent an extremely
frustrating 30 minutes or so trying to get cash from the neighborhood
Alternating Turing Machine.  And as if that wasn't enough, when I got back
to the office, my buddy William came in and forced me to remember the proof
that the second order theory of one's successor could be decided by an
Automated Teller Machine.  It's enough to drive a man MADD.
_______________________________________________________________________________
Russell G. Brown            | these opinions are reflected in full by /dev/null 

In article <54...@bu.edu.bu.edu> a...@cs.bu.edu (Al Thompson) writes:
>In article <10...@cbmvax.commodore.com> mks@cbmvax (Michael Sinz - CATS) writes:
>|In article <54...@bu.edu.bu.edu> a...@cs.bu.edu (Al Thompson) writes:
>[...]
>|>
>|>Time out!  The opcode is a number, the letters are a mnemonic device so
>|>the programmer can avoid remembering the number.  There is no necessary
>|>correlation between the number of letters in the mnemonic and the number
>|>it represents.
>|
>|Yes, however, due to memory limits, many of the assemblers were written
>|such that 3-bytes for the op-code with the 4th byte containing some data
>|on how to build that instruction.  It made the search/parse code really
>|simple.  (My PDP-11 assembler had that until Macro-11 came out and was
>|just tons better and needed a disk drive to work at reasonable rates)
>
>Sure, but that is an engineering decision.  But, there is no necessary
>correlation between the size of the mnemonic and the opcode.

I did not say that it was necessary.  I just said that one of the reasons
so many of the older machines had "Fixed-Length-Mnemonics" was due to the
way assemblers were designed, not due to the opcodes.

/----------------------------------------------------------------------\
|      /// Michael Sinz -- CATS/Amiga Software Engineer                |
|     ///  PHONE 215-431-9422  UUCP ( uunet | rutgers ) !cbmvax!mks    |
|    ///                                                               |
|\\\///          "I don't think so," said Ren'e Descartes.             |
| \XX/                    Just then, he vanished.                      |
\----------------------------------------------------------------------/

>>>>> On 2 Apr 90 13:00:27 GMT, rbr...@sjofn.cs.cornell.edu (Russell Brown) said:
RB> 	We obviously need a larger alphabet, since there are nowhere
RB> near enough three letter acronyms to go around.  After all, 26^3
RB> is only 17576, and that's not even enough to give a TLA to each
RB> VAX machine instruction, let alone cover all of the commercial and
RB> military applications.  Maybe we should follow U*ix's lead and
RB> make acronyms case sensitive.  That will buy us a factor of eight
RB> in the number of distinct TLAs.

This is a possible solution, but not the optimal one.  An alternate 
solution is to switch to ETLA (Enhanced TLA).  This provides a factor 
of twenty six improvement.  
--
------------------------------------------------------------------------------
zmacx07@uk.ac.ic.doc | sispe...@specialix.co.uk | ..!ukc!slxsys!cix!sispero
------------------------------------------------------------------------------
Can programming be liberated from the Von Neumann style? No FP, No Comment.
"The GNU Manifesto refers to all Software, not just Editors"    | (I'm the FSF)
					

scli...@phoenix.Princeton.EDU (Scott Charles Lindhurst) writes:
>6502 assembler mnemonics were all 3 letters long [...]
>(And the different mnemonics (and different
>opcodes, too) probably made writing assemblers easier.)

Sure did, although disassemblers where what really needed tight code.

jim frost
saber software
ji...@saber.com

The following is reproduced with no permission at all from 'Backbytes' -
the irreverent back page of the UK weekly 'Computing':

   As if they didn't have enough problems hard-pressed DEC distributors
   are having to work overtime keeping up with a torrent of new
   DECspeak. 

   At a recent thrash, one major distributor was overheared compaling to
   a fellow-sufferer, 'The trouble with DEC is they've got a fixation
   about TLAs - Three Letter Acronyms.  But I hear they're about to
   bring out an FLA.'

   'Four Letter Acronym?' asked his colleague, clearly an innocent soul. 

   'Sort of,' was the reply.  'It's a Field-upgradeable three Letter
   Acronym.'

Sam Wilson