[TUHS] sml/nj and unix/plan9

[TUHS] sml/nj and unix/plan9

[gdiaz]

[-- Attachment #1: Type: text/plain, Size: 399 bytes --]

hello

Was sml/nj part of UNIX at some point? was it considered as a language to use 
(proof tools may be)? 

I was wondering if there is any history in common between the two. I've been 
unable to find anything :-?, please share your stories! :-D

Is it true that the language was too slow to be generally useful? There seems 
to be commentaries along these lines on the internet.

thanks!
gabi




[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 839 bytes --]

Re: [TUHS] sml/nj and unix/plan9

[Andrew Koenig]

> Was sml/nj part of UNIX at some point? was it considered as a language to
use
> (proof tools may be)?
> 
> I was wondering if there is any history in common between the two. I've
been
> unable to find anything :-?, please share your stories! :-D
> 
> Is it true that the language was too slow to be generally useful? There
seems to be
> commentaries along these lines on the internet.

To my knowledge, sml/nj was never part of the Unix distribution, though it
was definitely available thereon (and also on SunOS). One of the main people
behind SML/NJ was Dave MacQueen, who was in the same general organization as
the Unix people.

As for sml/nj being too slow to be generally useful, Rob Pike (I think) once
wrote a desk-calculator program in C. I took that program and rewrote it in
sml/nj. Compared to the C version, it ran about twice as slowly and the
source code was about half the size. So no, I don't think sml/nj was slow.

Re: [TUHS] sml/nj and unix/plan9

[Dan Cross]

[-- Attachment #1: Type: text/plain, Size: 993 bytes --]

On Mon, May 4, 2020 at 8:49 AM <gdiaz@qswarm.com> wrote:

> Was sml/nj part of UNIX at some point? was it considered as a language to
> use
> (proof tools may be)?
>
> I was wondering if there is any history in common between the two. I've
> been
> unable to find anything :-?, please share your stories! :-D
>

There was certainly proximity, if not a direct connection.

Is it true that the language was too slow to be generally useful? There
> seems
> to be commentaries along these lines on the internet.
>

This question is difficult to answer. As a _langage_ there's little that
makes SML inherently slow; the MLton compiler does full-program
optimization with advanced optimizations and generates code that's pretty
performant. There are certainly other SML implementations that generate
slow code; MoscowML comes to mind: it generates a byte code that's not
known for speed. SML/NJ is pretty zippy, but I've never tried to write
anything performance-critical with it.

        - Dan C.

[-- Attachment #2: Type: text/html, Size: 1524 bytes --]

Re: [TUHS] sml/nj and unix/plan9

[A. P. Garcia]

[-- Attachment #1: Type: text/plain, Size: 592 bytes --]

On Mon, May 4, 2020, 8:48 AM <gdiaz@qswarm.com> wrote:

> hello
>
> Was sml/nj part of UNIX at some point? was it considered as a language to
> use
> (proof tools may be)?
>
> I was wondering if there is any history in common between the two. I've
> been
> unable to find anything :-?, please share your stories! :-D
>
> Is it true that the language was too slow to be generally useful? There
> seems
> to be commentaries along these lines on the internet.
>
> thanks!
> gabi
>
>
>
>
> If you Google Unix ML, there are two fairly  sizable papers on the topic
> near the top of the results...

[-- Attachment #2: Type: text/html, Size: 1153 bytes --]