23 Jul
2022
23 Jul
'22
4:57 a.m.
I've been looking into the history of the nl command lately, which has gotten me
curious as to what facilities folks have used at various points in UNIX history for line
numbering.
The earliest version of nl I've found is in System III, and it does not derive from
Research, PWB, or CB. Neither does it come from BSD, although BSD has the num command
which, according to the source commentary, aims to replicate the '#' behavior of
ex.
Were there any other facilities for printing back arbitrary lines from a file with line
numbers?
Also, would anyone happen to know if the above appearance of nl might have been from the
USG line given none of the others feature it? It neither seems to be in V8-V10. nl has
managed to find its way into the POSIX standard, so it definitely has some staying power
wherever it came from.
- Matt G.
9:55 a.m.
On Friday, July 22nd, 2022 at 10:56 PM, markus
schnalke <meillo(a)marmaro.de> wrote:
As I'm currently thinking a lot about ed, this comes to mind:
echo 23,42n | ed - file
The n rule is actually what lead me to studying nl. It's also supported in System
III but not V7, although ed's n does show up in V8-V10. The implementation looks
similar although integrated slightly differently. Probably from the same source though.
There's also an implementation in 4.4BSD-Lite, although plan old 4.4 with the
AT&T encumbered ed doesn't implement n.
So a rough timeline I can see for line number filters based on source code, old docs,
etc:
1977 - num and ex's '#' commands are introduced for 2BSD release
1980 - nl and ed's 'n' commands are introduced in System III
1980 - num's trail goes dark between 4 and 4.1c BSD
1982 - System V adopts ex, including '#'
1985 - Research has adopted ed with n by now as well as ex, never adopts nl
1992 - POSIX.2 is released, codifying nl, ed with n, and ex with #
1992 - nl appears in GNU fileutils
1994 - 4.4BSD-Lite ed includes n
1999 - NetBSD 1.4 adds nl
2002 - FreeBSD 4.5 adds nl
2005 - MacOSX 10.4 with text_cmds-47 adds nl
2013 - MINIX 3.2.1 adopts NetBSD nl
2014 - OpenBSD 5.5 adds nl
So all in all, num was an attempt that petered out, but # stayed in ex. System III and
4.4BSD-Lite ed derivatives implement n, and nl was introduced in System III for the
AT&T line and slowly cropped up elsewhere. System V picked up ex in 1982, allowing
nl, ed with n, and ex with # in the AT&T line. GNU had one out of the gate when POSIX
hit. BSDs were late to the party with regards to both ed with n as well as nl, with nl
only being adopted after the last Berkeley releases and subsequent fracturing.
So thus far it seems the earliest line numberer to see wide use was probably ex's #.
System III's nl became the ubiquitous tool for the job and was then slowly adopted
elsewhere. Ed rides a fine middle ground with the n command that saw wider adoption
earlier. I still have to wonder if there was something earlier though.
- Matt
1:01 p.m.
On Sat, Jul 23, 2022, 3:56 AM segaloco via TUHS <tuhs(a)tuhs.org> wrote:
[snip]
So all in all, num was an attempt that petered out, but # stayed in ex.
System III and 4.4BSD-Lite ed derivatives implement n, and nl was
introduced in System III for the AT&T line and slowly cropped up
elsewhere. System V picked up ex in 1982, allowing nl, ed with n, and ex
with # in the AT&T line. GNU had one out of the gate when POSIX hit. BSDs
were late to the party with regards to both ed with n as well as nl, with
nl only being adopted after the last Berkeley releases and subsequent
fracturing.
It may be worth noting that BSD had `cat -n` in 4BSD by October, 1980:
https://minnie.tuhs.org/cgi-bin/utree.pl?file=4BSD/usr/man/man1/cat.1
That may explain the relatively late incorporation of `nl` in, at least,
the BSD lineage.
- Dan C.
1:20 p.m.
On Sat, Jul 23, 2022 at 7:03 AM Dan Cross <crossd(a)gmail.com> wrote:
It may be worth noting that BSD had `cat -n` in 4BSD
by October, 1980:
https://minnie.tuhs.org/cgi-bin/utree.pl?file=4BSD/usr/man/man1/cat.1
That may explain the relatively late incorporation of `nl` in, at least,
the BSD lineage.
An obvious approach, which would leave no real traces in documentation,
would be:
$ awk '{print NR, $0}'
A more precise emulation would be more of a pain to type:
$ awk '{printf("%6d\t%s\n", NR, $0)}'
but perfectly usable in a script.
2 p.m.
On Sat, Jul 23, 2022 at 7:20 AM John Cowan <cowan(a)ccil.org> wrote:
On Sat, Jul 23, 2022 at 7:03 AM Dan Cross
<crossd(a)gmail.com> wrote:
Yes, but awk wasn't widely available until 7th edition. I imagine work
on it began before `num` in 2BSD, but few outside of Bell Labs would
have seen it prior to 1978 or so.
I wonder if the use-case was just sufficiently rare that no one felt like
building a special tool and it was just done on an ad-hoc basis, if
necessary.
- Dan C.
It may be worth noting that BSD had `cat -n` in
4BSD by October, 1980:
https://minnie.tuhs.org/cgi-bin/utree.pl?file=4BSD/usr/man/man1/cat.1
That may explain the relatively late incorporation of `nl` in, at least, the BSD lineage.
An obvious approach, which would leave no real traces in documentation, would be:
$ awk '{print NR, $0}'
A more precise emulation would be more of a pain to type:
$ awk '{printf("%6d\t%s\n", NR, $0)}'
but perfectly usable in a script. 
2:49 p.m.
I had a vague memory that pr could be made
to number lines, but a quick check of the 7/e
manual says no.
I expect Dan's right, and none of the 127 folks
felt much need to number lines on printouts
so nobody wrote the obvious simple tool.
Ironic, since the Unix PDP-11 used by the patent
licensing office (and I think shared with the
research group, and that was how their first
PDP-11 was justified and funded) happened
because the patent folks needed line-numbered
output and roff was easily modified to do that.
Maybe Doug or Ken or Steve has first-hand
memories.
Norman Wilson
Toronto ON
(on a train shuffling toward Buffalo)
3:20 p.m.
Very odd. I thought so too, but the 8th Edition manual says pr -n
prints in n columns, while the 9th and 10th say it numbers the lines.
No memory, if I ever knew, of what triggered that change.
-rob
On Sat, Jul 23, 2022 at 10:50 PM Norman Wilson <norman(a)oclsc.org> wrote:
I had a vague memory that pr could be made
to number lines, but a quick check of the 7/e
manual says no.
I expect Dan's right, and none of the 127 folks
felt much need to number lines on printouts
so nobody wrote the obvious simple tool.
Ironic, since the Unix PDP-11 used by the patent
licensing office (and I think shared with the
research group, and that was how their first
PDP-11 was justified and funded) happened
because the patent folks needed line-numbered
output and roff was easily modified to do that.
Maybe Doug or Ken or Steve has first-hand
memories.
Norman Wilson
Toronto ON
(on a train shuffling toward Buffalo)
3:36 p.m.
Hi Rob,
Very odd. I thought so too, but the 8th Edition manual
says pr -n
prints in n columns, while the 9th and 10th say it numbers the lines.
8th Edition's pr.c has -n and -x which both turn on line numbers.
The optional argument gives the width used for the line number.
https://minnie.tuhs.org/cgi-bin/utree.pl?file=V8/usr/src/cmd/pr.c
case 'n':
case 'x': /* retained for historical reasons */
++Lnumb;
if ((Numw = intopt(argv, &Nsepc)) <= 0)
Numw = NUMW;
if (Lnumb && C != EOF && (colno == 0 || Multi == 'a')) {
if (Page >= Fpage) {
putspace();
printf("%*ld", Numw, Buffer ?
Colpts[colno].c_lno++ : Lnumb);
Outpos += Numw;
put(Nsepc);
}
++Lnumb;
}
I think -n and -x are undocumented in the matching man page.
https://minnie.tuhs.org/cgi-bin/utree.pl?file=V8/usr/man/man1/pr.1
The -n you're referring to is an italic ‘n’, e.g. -3 for three columns.
.TP
.BI \- n
Produce
.IR n -column
output.
--
Cheers, Ralph.
4:33 p.m.
On Sat, Jul 23, 2022 at 9:21 AM Rob Pike <robpike(a)gmail.com> wrote:
Very odd. I thought so too, but the 8th Edition manual says pr -n
prints in n columns,
I'm almost certain that -n here means that -2 is 2 columns, -3 columns,
etc., and that there is no actual -n switch. In Posix as well as more
modern man pages this is spelled -COLUMNS: see man.cat-v.org.
while the 9th and 10th say it numbers the lines.
I think this may have come from BSD; it is now part of Posix.
24 Jul
24 Jul
9:45 p.m.
By 1984 (so some years after the formative stage in the era between 4.2BSD
and 4.3BSD), there
were a number of different programs on the net (USENET's source groups) to
pretty print things
on laser or line printers: grind, tgrind, vgrind, ete. These all had a
dozen different ways to gaudy up
the output (err, present the listing in a more aesthetically pleasing way):
Line numbers, line numbers
every 5 or 10 lines, line spacing tweaks, keywords as bold or italics,
frames around the output,
"two up" printing, various special behavior for functions (grey bands for
start of functions, function
definitions in bold, etc), as well as specialized options for assembler /
object renderings, etc. Most
were targeted at laser printers, but some were a better version of pr/lpr
for line printer output. Some
targeted postscript directly, while others used TeX, troff, etc in a
pipeline to gain some device
independence. Oh what a long way from pr -n these were. But at the time it
really helped me to
understand the power of the unix philosophy because these 'all in'
approaches were great until
you wanted to get a few steps beyond the beaten path and then it became
impossible (given the
great number of combinatorics for these options, many of the programs
produced odd results for
some combinations of options).
For day to day stuff, I hated these and used simpler options. But for
producing nice looking listings
for appendixes for papers / reports / assignments I'd written for school
they weren't half bad if
you avoided the worst of the gaudiness options. :)
Warner
On Sat, Jul 23, 2022 at 7:21 AM Rob Pike <robpike(a)gmail.com> wrote:
Very odd. I thought so too, but the 8th Edition manual
says pr -n
prints in n columns, while the 9th and 10th say it numbers the lines.
No memory, if I ever knew, of what triggered that change.
-rob
On Sat, Jul 23, 2022 at 10:50 PM Norman Wilson <norman(a)oclsc.org> wrote:
I had a vague memory that pr could be made
to number lines, but a quick check of the 7/e
manual says no.
I expect Dan's right, and none of the 127 folks
felt much need to number lines on printouts
so nobody wrote the obvious simple tool.
Ironic, since the Unix PDP-11 used by the patent
licensing office (and I think shared with the
research group, and that was how their first
PDP-11 was justified and funded) happened
because the patent folks needed line-numbered
output and roff was easily modified to do that.
Maybe Doug or Ken or Steve has first-hand
memories.
Norman Wilson
Toronto ON
(on a train shuffling toward Buffalo)
10:33 p.m.
All this discussion is making me incredibly glad I posed the question, as it does indeed
seem that before num and nl, line numbering was more of a "to each their own"
rather than there being a canonical best practice.
By the way, what inspired the whole quandary in the first place was doing some editing on
a V6 UNIX system and realizing I hadn't the foggiest idea of how to get the system to
spit back line numbers to remind myself what addresses to supply to various ed commands.
I suppose at the time using a context search was such an effective means of getting
ed's cursor where you wanted it that having a means to number lines wasn't
crucial to effectively working in ed, so long as one was good with their regex and
searching.
I'm realizing I might need to reach out to the maintainer of the nl manpage in
FreeBSD (and perhaps other OSs and distros) as the history section of the page posits that
nl was first introduced in SVR2, when in reality the earliest public offering was System
III, although the trail goes cold at that point, I can only assume it came from USG.
- Matt G.
11:04 p.m.
On Sun, Jul 24, 2022 at 2:33 PM segaloco <segaloco(a)protonmail.com> wrote:
All this discussion is making me incredibly glad I
posed the question, as
it does indeed seem that before num and nl, line numbering was more of a
"to each their own" rather than there being a canonical best practice.
By the way, what inspired the whole quandary in the first place was doing
some editing on a V6 UNIX system and realizing I hadn't the foggiest idea
of how to get the system to spit back line numbers to remind myself what
addresses to supply to various ed commands. I suppose at the time using a
context search was such an effective means of getting ed's cursor where you
wanted it that having a means to number lines wasn't crucial to effectively
working in ed, so long as one was good with their regex and searching.
I'm realizing I might need to reach out to the maintainer of the nl
manpage in FreeBSD (and perhaps other OSs and distros) as the history
section of the page posits that nl was first introduced in SVR2, when in
reality the earliest public offering was System III, although the trail
goes cold at that point, I can only assume it came from USG.
I can update it. Done.
https://cgit.freebsd.org/src/commit/?id=8a02ea1dbc139862456f488464072719341…
Warner
23 Jul
23 Jul
7:35 p.m.
Norman - I think your memory is fine.
pr -n (lower case) was an addition early on with pr +1 or pr +2 *etc* also
parsed. I'm going to guess that addition was either a local hack done by a
lot of people or maybe on the Harvard Tape and distributed from there. It
certainly worked that way in the CMU v6 and later v7. But I also remember
that I had to reprogram my fingers to pr -N (capital N) when I hit other
versions ??maybe AT&T Sys III??. I also remember that at CMU, tjk did not
like it / was not needed because he contended we already had num(1)
already on the EE systems (but not CS BTW). Ifwe had had shell aliases in
those days, I might have agreed with him. But given that it was not part
of v5, v6 or v7, that we only had a num(1) on the EE and Mellon systems and
not generally rippled to the other CMU UNIX systems around campus, I going
to take a WAG that Ted brought num(1) with him from UoMich (- which would
make sense - because the BSD history says Joy wrote it). So maybe
??probably?? the origin story for num(1) is that it came to UCB from Umich
when Bill and Ted were undergrads there.
FWIW: pr -n is what macOS does today.
ᐧ
On Sat, Jul 23, 2022 at 8:50 AM Norman Wilson <norman(a)oclsc.org> wrote:
I had a vague memory that pr could be made
to number lines, but a quick check of the 7/e
manual says no.
I expect Dan's right, and none of the 127 folks
felt much need to number lines on printouts
so nobody wrote the obvious simple tool.
Ironic, since the Unix PDP-11 used by the patent
licensing office (and I think shared with the
research group, and that was how their first
PDP-11 was justified and funded) happened
because the patent folks needed line-numbered
output and roff was easily modified to do that.
Maybe Doug or Ken or Steve has first-hand
memories.
Norman Wilson
Toronto ON
(on a train shuffling toward Buffalo)
8:51 p.m.
Phil Budne <phil(a)ultimate.com> suggests "grep -n ." to print numbered
lines.
That mostly works, but loses blank lines, although they are properly
counted. I typically use
awk '{print FNR ":" $0}'
instead, which reports all lines, regardless of contents.
-------------------------------------------------------------------------------
- Nelson H. F. Beebe Tel: +1 801 581 5254 -
- University of Utah -
- Department of Mathematics, 110 LCB Internet e-mail: beebe(a)math.utah.edu -
- 155 S 1400 E RM 233 beebe(a)acm.org beebe(a)computer.org -
- Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe/ -
-------------------------------------------------------------------------------
9:07 p.m.
Did you use "grep -n '.' "?
grep -n '^' matches blank lines, too.
Doug
On Sat, Jul 23, 2022 at 2:51 PM Nelson H. F. Beebe <beebe(a)math.utah.edu> wrote:
Phil Budne <phil(a)ultimate.com> suggests "grep -n ." to print numbered
lines.
That mostly works, but loses blank lines, although they are properly
counted. I typically use
awk '{print FNR ":" $0}'
instead, which reports all lines, regardless of contents.
-------------------------------------------------------------------------------
- Nelson H. F. Beebe Tel: +1 801 581 5254 -
- University of Utah -
- Department of Mathematics, 110 LCB Internet e-mail: beebe(a)math.utah.edu -
- 155 S 1400 E RM 233 beebe(a)acm.org beebe(a)computer.org -
- Salt Lake City, UT 84112-0090, USA URL: http://www.math.utah.edu/~beebe/ -
-------------------------------------------------------------------------------
24 Jul
24 Jul
9:02 p.m.
On Sat, Jul 23, 2022 at 08:00:10AM -0400, Dan Cross wrote:
My sed-fu is only rudimentary but after banging my head against
monitor a bit, I came up with this:
$ wc -l c.lisp
90 c.lisp
$ cat c.lisp | sed -e '{=;}' | sed -e 'N;s/\n/ /'
Seems to print what is needed. Now, just embed it in sh script. A
program in C could be longer, and required a compiler.
--
Regards,
Tomasz Rola
--
** A C programmer asked whether computer had Buddha's nature. **
** As the answer, master did "rm -rif" on the programmer's home **
** directory. And then the C programmer became enlightened... **
** **
** Tomasz Rola mailto:tomasz_rola@bigfoot.com **
On Sat, Jul 23, 2022 at 7:20 AM John Cowan
<cowan(a)ccil.org> wrote:
[...]
An obvious
approach, which would leave no real traces in documentation, would be:
$ awk '{print NR, $0}'
A more precise emulation would be more of a pain to type:
$ awk '{printf("%6d\t%s\n", NR, $0)}'
but perfectly usable in a script.
Yes, but awk wasn't widely available until 7th edition. I imagine work
on it began before `num` in 2BSD, but few outside of Bell Labs would
have seen it prior to 1978 or so.
I wonder if the use-case was just sufficiently rare that no one felt like
building a special tool and it was just done on an ad-hoc basis, if
necessary.
28 Jul
28 Jul
2:30 a.m.
On Sun, Jul 24, 2022 at 09:02:53PM +0200, Tomasz Rola wrote:
On Sat, Jul 23, 2022 at 08:00:10AM -0400, Dan Cross
wrote:
[...]
On Sat, Jul 23, 2022 at 7:20 AM John Cowan
<cowan(a)ccil.org> wrote:
[...]
> > An obvious approach, which would leave no real traces in documentation, would
be:
> >
> > $ awk '{print NR, $0}' > > $ awk '{printf("%6d\t%s\n",
NR, $0)}'
[...]
> Yes, but awk wasn't widely available until
7th edition. I imagine work
> on it began before `num` in 2BSD, but few outside of Bell Labs would
> have seen it prior to 1978 or so.
[...]
$ cat c.lisp | sed -e '{=;}' | sed -e
'N;s/\n/ /'
Ok, so maybe this is no longer very interesting subject, but I
recalled there was once a language named SNOBOL.
Wikipedia says, sed was written around 1974 [1] and about SNOBOL it
says, the language was being taught [2] in the late 1960s and early 1970s
in universities, so I guess it was quite popular. I wonder if it was
ported to Unix early enough to help before sed became available?
Anyway, I have got Phil Budne's implementation [3] and after usual
drill (configure, make, make install, stow) I could do this:
$ snobol4 -b -L ./numlines02.sno < numlines02.sno
1 lnum = 0
2 loop lnum = lnum + 1
3 output = lnum " " input :s(loop)
4 end
The first version did not work so well:
$ snobol4 -b -L ./numlines01.sno < numlines01.sno | head -8
1 lnum = 0
2 LOOP line = INPUT
3 * Here: detect EOF (how?) and jump to : END
4 lnum = lnum + 1
5 OUTPUT = lnum " " line : S(LOOP)
6 END
7 END
8 END
As I already have understood, even after reaching end of input, the
INPUT statement keeps returning last line read, counter keeps
increasing and last line is being written out for infinity. I have not
read any serious docs on SNOBOL yet, so I have no idea how to make the
'01' version work, but this is of rather small importance to the list.
However, since [2] claims SNOBOL was written in Bell Labs, just like
AWK and sed, so perhaps they could be using SNOBOL?
[1] https://en.wikipedia.org/wiki/Sed
[2] https://en.wikipedia.org/wiki/SNOBOL
[3] http://www.regressive.org/snobol4/
--
Regards,
Tomasz Rola
--
** A C programmer asked whether computer had Buddha's nature. **
** As the answer, master did "rm -rif" on the programmer's home **
** directory. And then the C programmer became enlightened... **
** **
** Tomasz Rola mailto:tomasz_rola@bigfoot.com **
3:03 a.m.
Anyway, I have got Phil Budne's implementation
C'est moi! SNOBOL came out of Bell Labs in Holmdel NJ.
There was a SNOBOL3 implementation in Unix 6th Edition days called "sno".
As far as I know Macro SNOBOL4 (that my CSNOBOL4 is a port of) never
was ported to the PDP-11 (just not enough address space), but there
was a proposal (at least) for a SNOBOL4 implementation for the '11
called ELFBOL.
Macro SPITBOL (a faster implementation of SNOBOL4) was available on
the Research Unix VAXen (Andrew Koenig did a C-like preprocessor
called SNOCONE -- SNOBOL with some sugar).
Phil
6:13 a.m.
New subject: SNOBOL and RATSNO
If memory serves, Ralph Griswold once told me that the four versions of SNOBOL,
culminating with SNOBOL4, took, respectively, a day, a week, a month, and year to
implement.
Related to SNOBOL, something I’ve wondered about for years is whether I may have been the
only regular user of RATSNO, which Dave Hanson created as a test of RATFOR’s
retargetability. I stumbled across RATSNO on the Icon V2 tape, and used it for quite a
bit of my undergraduate programming at NC State in 1980-81.
Phil Budne: Thanks for your CSNOBOL4 implementation! I’ve used it to show students
SNOBOL4 in a comparative languages class at the U of Arizona. (I was thinking your name
sounded familiar!)
On Jul 27, 2022, at 7:03 PM, Phil Budne
<phil(a)ultimate.com> wrote:
Anyway, I have got Phil Budne's
implementation
C'est moi! SNOBOL came out of Bell Labs in Holmdel NJ.
There was a SNOBOL3 implementation in Unix 6th Edition days called "sno".
As far as I know Macro SNOBOL4 (that my CSNOBOL4 is a port of) never
was ported to the PDP-11 (just not enough address space), but there
was a proposal (at least) for a SNOBOL4 implementation for the '11
called ELFBOL.
Macro SPITBOL (a faster implementation of SNOBOL4) was available on
the Research Unix VAXen (Andrew Koenig did a C-like preprocessor
called SNOCONE -- SNOBOL with some sugar).
Phil 
29 Jul
29 Jul
6:28 a.m.
New subject: SNOBOL and RATSNO
I did get SPITBOL to work past its expiry date on OS/360 :-) It was
dubbed as the "Superzap of the year" by one of my CompSci lecturers (Dr.
G.McMahon, UNSW).
The first couple of time-bombs were easy to find, but not so the rest
(long story).
Probably belongs over on COFF now...
-- Dave
7:07 a.m.
New subject: SNOBOL and RATSNO
On Wed, Jul 27, 2022 at 10:13:04PM -0600, William H. Mitchell wrote:
[...]
> Phil Budne: Thanks for your CSNOBOL4 implementation! I’ve used it to show students
SNOBOL4 in a comparative languages class at the U of Arizona. (I was thinking your name
sounded familiar!)
> > On Jul 27, 2022, at 7:03 PM, Phil Budne <phil(a)ultimate.com>
wrote:
>
> >> Anyway, I have got Phil Budne's implementation
>
> > C'est moi! SNOBOL came out of Bell Labs in Holmdel NJ.
> > There was a SNOBOL3 implementation in Unix 6th Edition days called
"sno".
[...]
Yes, I have had a look and it seems to be very nicely written
project. Oh, and there is plenty of Snobol4 code to look at, too...
Thank you.
--
Regards,
Tomasz Rola
--
** A C programmer asked whether computer had Buddha's nature. **
** As the answer, master did "rm -rif" on the programmer's home **
** directory. And then the C programmer became enlightened... **
** **
** Tomasz Rola mailto:tomasz_rola@bigfoot.com **
9 Aug
9 Aug
7:12 a.m.
New subject: SNOBOL and RATSNO
On 7/29/2022 1:07 AM, Tomasz Rola wrote:
On Wed, Jul 27, 2022 at 10:13:04PM -0600, William H.
Mitchell wrote:
[...]
Speaking of SNOBOL4, I typed up the SNOBOL code from the NRL Report 7948
(1975) titled "Automatic Translation of English Text to Phonetics by
Means of Letter-to-Sound Rules" by Honey Sue Elovitz, Rodney W. Johnson,
Astrid McHugh and John E. Shore, and made some minor modifications to
make it work properly with the windows/catspaw version of snobol/spitbol.
It might not be necessary to make those changes at all, with Phil's
version, I'll need to try that!
I have both the patched and unpatched versions at
https://github.com/Lord-Nightmare/NRL_TextToPhonemes and it does behave
correctly/matches the paper (at least the patched version does).
I recently (within the past month) discovered another later port of the
NRL ruleset from 1978 as part of Peter B. Maggs' ANGLOPHONE package for
S-100 systems, intended for use with the Computalker CT-1 speech
synthesis S-100 card. Apparently Rodney W. Johnson had continued
developing the rules even after the 1975/1976 publications of the NRL
report and the IEEE ITASSP version of said report, and I haven't updated
the bibliography on the github readme yet.
Jonathan G.
--
Jonathan Gevaryahu AKA Lord Nightmare
jgevaryahu(a)gmail.com
jgevaryahu(a)hotmail.com
Phil Budne: Thanks for your CSNOBOL4
implementation! I’ve used it to show students SNOBOL4 in a comparative languages class at
the U of Arizona. (I was thinking your name sounded familiar!)
> On Jul 27, 2022, at 7:03 PM, Phil Budne <phil(a)ultimate.com> wrote:
>
>> Anyway, I have got Phil Budne's implementation
> C'est moi! SNOBOL came out of Bell Labs in Holmdel NJ.
> There was a SNOBOL3 implementation in Unix 6th Edition days called "sno".
[...]
Yes, I have had a look and it seems to be very nicely written
project. Oh, and there is plenty of Snobol4 code to look at, too...
Thank you.
8:11 a.m.
New subject: SNOBOL and RATSNO
We're probably well off topic now but...
Many years ago I ran into Bob Dewar on a visit to Cambridge University and
we got to talking. He said that the original implementation of SPITBOL, for
the System/360, was in assembler (of course), and written by him and
Belcher (?). The story he told was that they wrote it all down first, put
it on punch cards, and sent it to the IBM machine. The next day they got
back a listing with a bunch of errors. They iterated. By the fourth
round—fifth day—they had a working SPITBOL.
I still marvel at the productivity and precision of his generation of
programmers.
-rob
On Tue, Aug 9, 2022 at 3:13 PM Jonathan Gevaryahu <jgevaryahu(a)hotmail.com>
wrote:
On 7/29/2022 1:07 AM, Tomasz Rola wrote:
On Wed, Jul 27, 2022 at 10:13:04PM -0600, William
H. Mitchell wrote:
[...]
> Phil Budne: Thanks for your CSNOBOL4 implementation! I’ve used it to
show
students SNOBOL4 in a comparative languages class at the U of
Arizona. (I was thinking your name sounded familiar!)
>
>> On Jul 27, 2022, at 7:03 PM, Phil Budne <phil(a)ultimate.com> wrote:
>>
>>> Anyway, I have got Phil Budne's implementation
>> C'est moi! SNOBOL came out of Bell Labs in Holmdel NJ.
>> There was a SNOBOL3 implementation in Unix 6th Edition days called
"sno".
[...]
Yes, I have had a look and it seems to be very nicely written
project. Oh, and there is plenty of Snobol4 code to look at, too...
Thank you.
Speaking of SNOBOL4, I typed up the SNOBOL code from the NRL Report 7948
(1975) titled "Automatic Translation of English Text to Phonetics by
Means of Letter-to-Sound Rules" by Honey Sue Elovitz, Rodney W. Johnson,
Astrid McHugh and John E. Shore, and made some minor modifications to
make it work properly with the windows/catspaw version of snobol/spitbol.
It might not be necessary to make those changes at all, with Phil's
version, I'll need to try that!
I have both the patched and unpatched versions at
https://github.com/Lord-Nightmare/NRL_TextToPhonemes and it does behave
correctly/matches the paper (at least the patched version does).
I recently (within the past month) discovered another later port of the
NRL ruleset from 1978 as part of Peter B. Maggs' ANGLOPHONE package for
S-100 systems, intended for use with the Computalker CT-1 speech
synthesis S-100 card. Apparently Rodney W. Johnson had continued
developing the rules even after the 1975/1976 publications of the NRL
report and the IEEE ITASSP version of said report, and I haven't updated
the bibliography on the github readme yet.
Jonathan G.
--
Jonathan Gevaryahu AKA Lord Nightmare
jgevaryahu(a)gmail.com
jgevaryahu(a)hotmail.com
3:34 p.m.
New subject: SNOBOL and RATSNO
On Tue, Aug 9, 2022 at 2:12 AM Rob Pike <robpike(a)gmail.com> wrote:
I still marvel at the productivity and precision of his generation of
programmers.
Amen.
When I was first learning the ins and outs of the implementation of the
York/APL System for the 360, I was regaled with similar stories and hoped
that I could measure up to their standards.
ᐧ
5:15 p.m.
New subject: SNOBOL and RATSNO
rob, clem:
has there been a shift in ability? or is this more likely a sampling bias
(because there were so many fewer programmers then)?
On Aug 9, 2022, at 6:34 AM, Clem Cole
<clemc(a)ccc.com> wrote:
On Tue, Aug 9, 2022 at 2:12 AM Rob Pike <robpike(a)gmail.com
<mailto:robpike@gmail.com>> wrote:
I still marvel at the productivity and precision of his generation of programmers.
Amen.
8:26 p.m.
New subject: SNOBOL and RATSNO
On Tue, Aug 9, 2022 at 11:15 AM Andrew Hume <andrew(a)humeweb.com> wrote:
rob, clem:
has there been a shift in ability? or is this more likely a sampling bias
(because there were so many fewer programmers then)?
Certainly part of it. But I think was more likely the type of person than
the number. In those days because the tools and use of them required
extreme precision, only the precise found their way.
What's the Pixar Incredible's line: *If everyone is super, no one will
be.*
ᐧ
ᐧ
8:52 p.m.
New subject: SNOBOL and RATSNO
I'll confess: I was never very good at bench checking batch programs,
but only had at most a handful of assignments in college: generally
cycles were cheap on time-sharing systems and I quickly adapted to
interactive debugging.
Over time (with embedded systems in networking gear and other
applications), that wasn't possible and the new skill I admired was
being able to add effective logging to diagnose problems. And since most
of those systems were some combination of
real-time/multiprocessor/multithreaded, it was generally not possible to
deterministically repeat the sequence.
On 8/9/22 11:15 AM, Andrew Hume wrote:
rob, clem:
has there been a shift in ability? or is this more likely a sampling bias
(because there were so many fewer programmers then)?
On Aug 9, 2022, at 6:34 AM, Clem Cole
<clemc(a)ccc.com> wrote:
On Tue, Aug 9, 2022 at 2:12 AM Rob Pike <robpike(a)gmail.com> wrote:
I still marvel at the productivity and precision of his
generation of programmers.
Amen.
11:25 p.m.
New subject: SNOBOL and RATSNO
I think both factors were relevant, as well as the preciousness and pace of
it all, as discussed by others here.
-rob
On Wed, Aug 10, 2022 at 1:15 AM Andrew Hume <andrew(a)humeweb.com> wrote:
rob, clem:
has there been a shift in ability? or is this more likely a sampling bias
(because there were so many fewer programmers then)?
On Aug 9, 2022, at 6:34 AM, Clem Cole <clemc(a)ccc.com> wrote:
On Tue, Aug 9, 2022 at 2:12 AM Rob Pike <robpike(a)gmail.com> wrote:
I still marvel at the productivity and precision of his generation of
programmers.
Amen.
5:39 p.m.
New subject: SNOBOL and RATSNO
On Tue, Aug 9, 2022 at 9:36 AM Clem Cole <clemc(a)ccc.com> wrote:
On Tue, Aug 9, 2022 at 2:12 AM Rob Pike <robpike(a)gmail.com> wrote:
If you don't know about Mel who wrote a cheating blackjack game for a drum
memory machine, read https://www.cs.utah.edu/~elb/folklore/mel.html And
the postscript.
I still marvel at the productivity and precision of his generation of
programmers.
Amen.
When I was first learning the ins and outs of the implementation of the
York/APL System for the 360, I was regaled with similar stories and hoped
that I could measure up to their standards.
ᐧ
3:56 p.m.
New subject: SNOBOL and RATSNO
On Tue, Aug 09, 2022 at 04:11:38PM +1000, Rob Pike wrote:
We're probably well off topic now but...
Many years ago I ran into Bob Dewar on a visit to Cambridge University and
we got to talking. He said that the original implementation of SPITBOL, for
the System/360, was in assembler (of course), and written by him and
Belcher (?). The story he told was that they wrote it all down first, put
it on punch cards, and sent it to the IBM machine. The next day they got
back a listing with a bunch of errors. They iterated. By the fourth
round???fifth day???they had a working SPITBOL.
I still marvel at the productivity and precision of his generation of
programmers.
I had the same reaction to pic(1). You could look at the code and "see"
what it was doing. I've always believed that pic was so well designed
because it took a day to get the print out (back then), so you had to
have a language where you could see what it was doing.
6:45 p.m.
New subject: SNOBOL and RATSNO
A cool thing you buy in those days was the full source code for SPITBOL, on microfiche. I
bought a copy and a number of my friends did, too. It was really beautiful code.
On Aug 8, 2022, at 11:11 PM, Rob Pike
<robpike(a)gmail.com> wrote:
We're probably well off topic now but...
Many years ago I ran into Bob Dewar on a visit to Cambridge University and we got to
talking. He said that the original implementation of SPITBOL, for the System/360, was in
assembler (of course), and written by him and Belcher (?). The story he told was that they
wrote it all down first, put it on punch cards, and sent it to the IBM machine. The next
day they got back a listing with a bunch of errors. They iterated. By the fourth
round—fifth day—they had a working SPITBOL.
I still marvel at the productivity and precision of his generation of programmers.
-rob
29 Jul
29 Jul
2:22 a.m.
New subject: SNOBOL and progeny [Was: Re: Re: Line Numbers Before SysIII nl? BSD num?]
On 2022-07-27 21:03, Phil Budne wrote:
I recall seeing ICEBOL card packs at the U. of Toronto Computing Centre
decades ago but I know nothing more.
N.
Anyway, I have
got Phil Budne's implementation
C'est moi! SNOBOL came out of Bell Labs
in Holmdel NJ.
There was a SNOBOL3 implementation in Unix 6th Edition days called "sno".
As far as I know Macro SNOBOL4 (that my CSNOBOL4 is a port of) never
was ported to the PDP-11 (just not enough address space), but there
was a proposal (at least) for a SNOBOL4 implementation for the '11
called ELFBOL.
Macro SPITBOL (a faster implementation of SNOBOL4) was available on
the Research Unix VAXen (Andrew Koenig did a C-like preprocessor
called SNOCONE -- SNOBOL with some sugar).
Phil 
7:01 a.m.
New subject: SNOBOL and progeny [Was: Re: Re: Line Numbers Before SysIII nl? BSD num?]
A few experiences as SNOBOL/Icon enthusiast, not expert, ending with limited tie back to
Unix:
o Introduced to SNOBOL4 in introductory programming languages course summer 1971 U.T.
Austin
o ca 1976 used SNOBOL4 on VM/370 to build Fortran to PL/I translator "The elapsed
time between beginning work on the translator and getting a running PL/I version of APLOMB
was approximately two weeks, and this achievement was a great relief to those who
anticipated a much, much larger effort."
https://technologists.com/sauer/The_Evolution_of_the_Research_Queueing_Pack…
<https://technologists.com/sauer/The_Evolution_of_the_Research_Queueing_Package.pdf>
o ca 1985 (with minimal effort) modified/built Icon to run on PC/IX, began emailing with
Ralph Griswold
o Discovered IBM colleague Viktors Berstis and his SNOBOL advocacy/expertise
(http://berstis.com/ <http://berstis.com/>)
o ca 1986 "With help from an ISC colleague, I created a PL.8 to C translator using
Icon, that was used to facilitate some of the rewriting. [of the RT/PC VRM]"
https://notes.technologists.com/notes/2017/03/08/lets-start-at-the-very-beg…
<https://notes.technologists.com/notes/2017/03/08/lets-start-at-the-very-beginning-801-romp-rtpc-aix-versions/>
Charlie
--
voice: +1.512.784.7526 e-mail: sauer(a)technologists.com
<mailto:sauer@technologists.com>
fax: +1.512.346.5240 web: https://technologists.com/sauer/
<http://technologists.com/sauer/>
Facebook/Google/Skype/Twitter: CharlesHSauer
4:07 p.m.
New subject: SNOBOL and progeny [Was: Re: Re: Line Numbers Before SysIII nl? BSD num?]
On Fri, Jul 29, 2022 at 1:02 AM Charles H. Sauer <sauer(a)technologists.com>
wrote:
A few experiences as SNOBOL/Icon enthusiast, not
expert, ending with
limited tie back to Unix:
o Introduced to SNOBOL4 in introductory programming languages course
summer 1971 U.T. Austin
o ca 1976 used SNOBOL4 on VM/370 to build Fortran to PL/I translator "The
elapsed time between beginning work on the translator and getting a running
PL/I version of APLOMB was approximately two weeks, and this achievement
was a great relief to those who anticipated a much, much larger effort."
https://technologists.com/sauer/The_Evolution_of_the_Research_Queueing_Pack…
o ca 1985 (with minimal effort) modified/built Icon to run on PC/IX, began
emailing with Ralph Griswold
o Discovered IBM colleague Viktors Berstis and his SNOBOL
advocacy/expertise (http://berstis.com/)
o ca 1986 "With help from an ISC colleague, I created a PL.8 to C
translator using Icon, that was used to facilitate some of the rewriting.
[of the RT/PC VRM]"
https://notes.technologists.com/notes/2017/03/08/lets-start-at-the-very-beg…
Let us not forget Icon <https://www2.cs.arizona.edu/icon/>, Ralph
Griswold's own successor to Snobol. Well-structured, portable to Posix and
Windows, "Prolog from another point of view" (the Icon implementation is
very much like the WAM).
5:37 p.m.
New subject: SNOBOL and progeny [Was: Re: Re: Line Numbers Before SysIII nl? BSD num?]
SNOBOL influenced Dick Haight's bs(1) programming language which
includes some SNOBOL features, a replacement, written in C, for Ken
Thompson's bas(1) BASIC implementation, written in assembly language.
bs (programming language)
https://en.wikipedia.org/wiki/Bs_(programming_language)
BS: a mysterious Unix programming language!
https://www.youtube.com/watch?v=ELICIa3L22o
Although I know of no applications written in bs(1), it remains to
this day in licensed System V Unix variants such as HP-UX and AIX.
Peace,
Dave
On Thu, Jul 28, 2022 at 7:22 PM Stuff Received <stuff(a)riddermarkfarm.ca> wrote:
On 2022-07-27 21:03, Phil Budne wrote:
I recall seeing ICEBOL card packs at the U. of Toronto Computing Centre
decades ago but I know nothing more.
N.
--
dave(a)plonka.us http://www.cs.wisc.edu/~plonka/
Anyway, I
have got Phil Budne's implementation
C'est moi! SNOBOL came out of Bell
Labs in Holmdel NJ.
There was a SNOBOL3 implementation in Unix 6th Edition days called "sno".
As far as I know Macro SNOBOL4 (that my CSNOBOL4 is a port of) never
was ported to the PDP-11 (just not enough address space), but there
was a proposal (at least) for a SNOBOL4 implementation for the '11
called ELFBOL.
Macro SPITBOL (a faster implementation of SNOBOL4) was available on
the Research Unix VAXen (Andrew Koenig did a C-like preprocessor
called SNOCONE -- SNOBOL with some sugar).
Phil
831
days inactive
848
days old
36 comments
23 participants
participants (23)
-
Andrew Hume -
Charles H. Sauer -
Clem Cole -
Dan Cross -
Dave Horsfall -
Dave Plonka -
Douglas McIlroy -
John Cowan -
Jonathan Gevaryahu -
Larry McVoy -
markus schnalke -
Nelson H. F. Beebe -
Norman Wilson -
Phil Budne -
Ralph Corderoy -
Richard Salz -
Rob Pike -
segaloco -
Stuff Received -
Tom Teixeira -
Tomasz Rola -
Warner Losh -
William H. Mitchell