From Newsgroup: alt.folklore.computers

On 2026-01-07, Charlie Gibbs <cgibbs@kltpzyxm.invalid> wrote:

A friend once wrote an 8080 cross-assembler in COBOL.
It ran rings around Univac's official cross-assembler -
which was written in FORTRAN.

So long as it did not include a macro facility, that actually makes
sense. Assembly code tended to have fixed column layout:
- labels starting in column 1
- opcode in column 9
- operands in column 17

One input line makes one instruction.

But trying to parse free-form text and do macro expansions with
string substitutions ... disaster in COBOL. Hard enough in FORTRAN.

When I went from academia to a small engineering firm in 1975,
I was used to writing my documentation in Univac DOC on the
university's mailframe, and hated having to go through a secretary
with a typewriter, who did not know the technical subject matter.

So I wrote a slightly modified re-implementation of DOC in
RSX-11M Fortran on our PDP-11 systems. That was fun, and my
colleagues liked it. The documents looked very good when printed
on a Diablo SpinWriter. A couple of years later, our firm became
the local representatives of Wang Labs, just as they introduced
the Wang WPS office systems, and we switched to that.

I think the magtapes with the source code from back then were
lost when I changed jobs around 1990. (Actually, the tapes may
still be in a box somewhere, but I don't know where I'd find
a tape drive where I could load them to see if they are
recoverable after 30+ years in garage and mini-storage.)
--
Lars Poulsen - an old geek in Santa Barbara, California
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On Wed, 7 Jan 2026 14:11:10 -0000 (UTC), Lars Poulsen wrote:

But trying to parse free-form text and do macro expansions with
string substitutions ... disaster in COBOL. Hard enough in FORTRAN.

Hard to see the point in an assembler without such features, though.
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On 1/7/26 13:21, Lawrence D’Oliveiro wrote:

On Wed, 7 Jan 2026 14:11:10 -0000 (UTC), Lars Poulsen wrote:

But trying to parse free-form text and do macro expansions with
string substitutions ... disaster in COBOL. Hard enough in FORTRAN.

Hard to see the point in an assembler without such features, though.

I did a COBOL program to do string substitutions. The idea was that it
read a COBOL program written possibly by a blind hacker and substituted variable names with longer, standardized ones.
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On Wed, 7 Jan 2026 19:16:03 -0700, Peter Flass wrote:

I did a COBOL program to do string substitutions. The idea was that it
read a COBOL program written possibly by a blind hacker and substituted variable names with longer, standardized ones.

Did it understand the rules of IN-scoping?
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On Wed, 7 Jan 2026 20:21:19 -0000 (UTC), Lawrence D’Oliveiro wrote:

On Wed, 7 Jan 2026 14:11:10 -0000 (UTC), Lars Poulsen wrote:

But trying to parse free-form text and do macro expansions with string
substitutions ... disaster in COBOL. Hard enough in FORTRAN.

Hard to see the point in an assembler without such features, though.

When the company balked at an expensive MCS-8048 cross assembler for the
VAX I wrote one that ran on CP/M. There was no point in such features.
When you're dealing with 2K of EEPROM and 128 bytes of RAM you're close to
the bone.


--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On 1/7/26 19:24, Lawrence D’Oliveiro wrote:

On Wed, 7 Jan 2026 19:16:03 -0700, Peter Flass wrote:

I did a COBOL program to do string substitutions. The idea was that it
read a COBOL program written possibly by a blind hacker and substituted
variable names with longer, standardized ones.

Did it understand the rules of IN-scoping?

COBOL had no concept of "scope" back then.
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On Thu, 8 Jan 2026 07:41:03 -0700, Peter Flass wrote:

On 1/7/26 19:24, Lawrence D’Oliveiro wrote:

On Wed, 7 Jan 2026 19:16:03 -0700, Peter Flass wrote:

I did a COBOL program to do string substitutions. The idea was that it
read a COBOL program written possibly by a blind hacker and substituted
variable names with longer, standardized ones.

Did it understand the rules of IN-scoping?

COBOL had no concept of "scope" back then.

I was referring to this:

8.4.1.1 Qualification

Qualification is used to allow unique reference of user names.
Qualification is the specification of superordinate names from the
hierarchy to which a user-defined name belongs. The superordinate
names are called qualifiers. Identical user-defined names may be
specified in a source unit; however, uniqueness shall be established
through qualification for each user- defined name explicitly
referenced, except as specified in rules 2 through 6. All available
qualifiers need not be specified so long as uniqueness is established.
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

Lars Poulsen <lars@beagle-ears.com> writes:

But trying to parse free-form text and do macro expansions with
string substitutions ... disaster in COBOL. Hard enough in FORTRAN.

Using "OCCURS DEPENDING ON" COBOL easily processes variable length,
variably located strings.

Not much of a disaster IMO.
--
Dan Espen
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On Fri, 09 Jan 2026 14:22:34 -0500, Dan Espen wrote:

Using "OCCURS DEPENDING ON" COBOL easily processes variable length,
variably located strings.

Up to a limit, always, e.g.

OCCURS [ integer-1 TO ] integer-2 TIMES [ DEPENDING ON data-name-1 ]
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

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

On Fri, 09 Jan 2026 14:22:34 -0500, Dan Espen wrote:

Using "OCCURS DEPENDING ON" COBOL easily processes variable length,
variably located strings.

Up to a limit, always, e.g.

OCCURS [ integer-1 TO ] integer-2 TIMES [ DEPENDING ON data-name-1 ]

Well beyond any reasonable limit.
DATA-NAME-1 can be the largest number format the compiler supports and INTEGER-2 the max value.
--
Dan Espen
--- Synchronet 3.21b-Linux NewsLink 1.2

From Newsgroup: alt.folklore.computers

On Sat, 10 Jan 2026 12:18:13 -0500, Dan Espen wrote:

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

On Fri, 09 Jan 2026 14:22:34 -0500, Dan Espen wrote:

Using "OCCURS DEPENDING ON" COBOL easily processes variable
length, variably located strings.

Up to a limit, always, e.g.

OCCURS [ integer-1 TO ] integer-2 TIMES [ DEPENDING ON data-name-1 ]

Well beyond any reasonable limit.

I think you’ll find that the implementation reserves space for the
maximum specified value. So this is not dynamic allocation, by any
means.
--- Synchronet 3.21b-Linux NewsLink 1.2