Staring a while at the HTTPS client example in Gambit-Scheme's manual[0]
I could minimise and morph it back to plain HTTP and then to NNTP and
NNTPS examples.

#+begin_example
(let ( (d (open-tcp-client (list address: "news.eternal-september.org"
port-number: 119
eol-encoding: 'cr-lf))) )
(print port: d "article <rrej7g$72d$1@dont-email.me>\nquit\n.\n")
(force-output d)
(print (read-line d #f)) )
#+end_example

#+begin_example
(let ( (d (open-tcp-client (list address: "news.eternal-september.org"
port-number: 563
eol-encoding: 'cr-lf
tls-context: (make-tls-context)))) )
(print port: d "article <rrej7g$72d$1@dont-email.me>\nquit\n.\n")
(force-output d)
(print (read-line d #f)) )
#+end_example

But between those HTTP and NNTP experiments I tried to access Gemini
with a similar minimised example and it stayed uncooperative. I'm still scratching my head about that, but currently being more inet-rested
(nice char swap / dash added later) in NNTP(S), the Gemini problem boils
down to the curiosity level instead of being essential. It may come
back somewhen later to bite me again, so if answers find me earlier than
that, I might save some time collecting client and server code examples
for (not only Smallnet) protocols (and not only in Scheme).

____________

[0]: https://gambitscheme.org/latest/manual
--
Trust me, I know what I'm doing...


;-D
--- Synchronet 3.19b-Linux NewsLink 1.113

yeti <yeti@tilde.institute> writes:

Staring a while at the HTTPS client example in Gambit-Scheme's manual[0]
I could minimise and morph it back to plain HTTP and then to NNTP and
NNTPS examples.

#+begin_example
(let ( (d (open-tcp-client (list address: "news.eternal-september.org"
port-number: 119
eol-encoding: 'cr-lf))) )
(print port: d "article <rrej7g$72d$1@dont-email.me>\nquit\n.\n")
(force-output d)
(print (read-line d #f)) )
#+end_example

#+begin_example
(let ( (d (open-tcp-client (list address: "news.eternal-september.org"
port-number: 563
eol-encoding: 'cr-lf
tls-context: (make-tls-context)))) )
(print port: d "article <rrej7g$72d$1@dont-email.me>\nquit\n.\n")
(force-output d)
(print (read-line d #f)) )
#+end_example

But between those HTTP and NNTP experiments I tried to access Gemini
with a similar minimised example and it stayed uncooperative.

You're saying you managed to access the NNTP server in both ways above,
but the Gemini protocol somehow gives you a problem? Are you able to
talk to the Gemini protocol with the command below?

openssl s_client -connect host.name:1234
--- Synchronet 3.19b-Linux NewsLink 1.113

Yipyip, I can access Gemini via GNUTLS-CLI, NC, Ncat.

I have to switch some clients to not refuse "insecure" certs (Gemini
often uses self signed sigs (and then trust on first use by design)).
This may be the rough edge and I do not see an option to handle that via
the OpenSSL options Gambit makes accessible.

#+begin_src sh
{ sleep 1; printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' ; } \
| gnutls-cli --insecure geminiprotocol.net:1965
#+end_src

#+begin_src sh
printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' \
| nc -c -T noverify geminiprotocol.net 1965
#+end_src

#+begin_src sh
printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' \
| ncat --ssl geminiprotocol.net 1965
#+end_src

Ncat seems tolerant by default.


It still may be something totally unrelated I'm still overlooking.


/!\ Note to $SELF: Add an OpenSSL snippet to my collection.
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

yeti <yeti@tilde.institute> writes:

[...]

#+begin_src sh
printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' \
| ncat --ssl geminiprotocol.net 1965
#+end_src

Ncat seems tolerant by default.

Meaning it accepts self-signed but if you pass an option --dont-do-it,
it doesn't? I like that: insecure by default. Why don't we let the
fearful people read the manual? Why must always the brave that need to
learn how to turn off all the bureaucracy?
--- Synchronet 3.19b-Linux NewsLink 1.113

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

Ncat seems tolerant by default.

Meaning it accepts self-signed but if you pass an option --dont-do-it,
it doesn't?

The man-page includes:

------------------------------------------------------------------------
--ssl-verify Verify trust and domain name of certificates ------------------------------------------------------------------------

Seems it shall be used instead of `--ssl`.

------------------------------------------------------------------------
--ssl-verify (Verify server certificates)
In client mode, --ssl-verify is like --ssl except that it
also requires verification of the server certificate. Ncat
comes with a default set of trusted certificates in the file
ca-bundle.crt. Some operating systems provide a default list
of trusted certificates; these will also be used if
available. Use --ssl-trustfile to give a custom list. Use -v
one or more times to get details about verification failures.
Ncat does not check for revoked certificates.

This option has no effect in server mode. ------------------------------------------------------------------------

Meaning it accepts self-signed but if you pass an option --dont-do-it,
it doesn't?

I'm not sure I grok Ncat's behaviour:

#+header: :prologue exec 2>&1 :epilogue true :results output :wrap example #+begin_src sh
printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' \
| ncat --ssl geminiprotocol.net 1965
#+end_src

#+RESULTS:
#+begin_example
20 text/gemini
# Project Gemini specifications

In late March 2024, the official Gemini specification was refactored into two separate documents, one which addresses only the network protocol and one which addresses only the markup format. Both specifications are intended to be more formal and less ambiguous than the combined document which historical served as a "speculative specification".

protocol-specification.gmi Gemini network protocol specification gemtext-specification.gmi Gemini hypertext format ("gemtext") specification

The old speculative specification, which used to live at the URL you're at right now, will evolve instead into a "technical overview". It's more detailed than the Project Gemini FAQ, and is aimed at geeks, not a general audience, but it stops short of covering every corner case and defining things in down-to-the-byte detail, for which you should see the appropriate specification above. The overview serves as a good and highly readable introduction to the technologies behind Gemini. You can probably still use it to write a client which works well enough, often enough to be genuinely useful.

tech-overview.gmi Project Gemini technical overview

If you've arrived at this page by following a link or a bookmark, please consider updating it, or asking for it to be updated, to point to the technical overview if you think that's more approprite for the intended audience than this explanatory page.
#+end_example

#+header: :prologue exec 2>&1 :epilogue true :results output :wrap example #+begin_src sh
printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' \
| ncat --ssl-verify geminiprotocol.net 1965
#+end_src

#+RESULTS:
#+begin_example
Ncat: TIMEOUT.
#+end_example

I like that: insecure by default. Why don't we let the fearful people
read the manual?

No idea.

Why must always the brave that need to learn how to turn off all the bureaucracy?

I'm not brave. I just think forcing TLS into Gemini is BS, like with
every other protocol. The outer tunnel to a server should be an
orthogonal dimension of the parameters of the connection and the user
should be deciding which one to use, including NONE as a valid
alternative. Stacking protocols should be modular instead of being
glued into some protocols and not into others.
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

yeti <yeti@tilde.institute> writes:

#+header: :prologue exec 2>&1 :epilogue true :results output :wrap example #+begin_src sh

printf 'gemini://geminiprotocol.net/docs/specification.gmi\r\n' \
| ncat --ssl-verify geminiprotocol.net 1965
#+end_src

#+RESULTS:

#+begin_example
Ncat: TIMEOUT.
#+end_example

I've zero idea why there are blank lines after "#+begin_src sh" and "#+RESULTS:". That needs a closer look somewhen... later... mayhaps...
--
|rom The Future. +++ Breaking News From The Future. +++ Breaking News F|
| The USoA are switching to the binary number system because |
| having more than 1+1 distinct digits is far too woke. |
|+ #MABA + #makeAmericaBinaryAgain + #USA + #USoA + #woke + #MABA + #ma|
--- Synchronet 3.19b-Linux NewsLink 1.113

Meanwhile that TIMEOUT has turned into a different error and I have no
idea why or what happened before.

Having found some Gemini servers I can access even with "--ssl-verify",
I'm fairly convinced now that Gambit just needs an option too to
dis-/arm that check on demand.

I could drop a question about this via Gwene into Gambit's ML:

| Newsgroups: gmane.lisp.scheme.gambit
| To: gambit-list@iro.umontreal.ca

Fingers X-ed... ,,X_

I'd like to continue using Gambit, but if I need to switch the Scheme
compiler to continue playing with those protocols, I will do exactly
that.

What I do not want is switching to Lisp. I'd prefer to only use one
Lisp and this shall be Emacs-Lisp and that one is really bad at network
access and even worse for making stand alone tools to run without Emacs.

Similarly for Scheme. My experiences with Debian and Ubuntu told me
long ago that very similar, but not identical is a bigger problem than
big differences, e.g. using two more distant Unix relatives side by
side.
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

Sometimes one just needs to be brave enough to ask, but that's a real
hurdle for me, especially in the context of contexts/languages where I
still have the feeling of not even having reached preschool level.

The ML quickly gave a reaction and a hint how to patch it. What's
really needed would be an option to do that at runtime and especially
per socket/connection, but it is a start and shows what should be
touched.

It's an issue now: <https://github.com/gambit/gambit/issues/904>
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

Sometimes one just needs to be brave enough to ask, but that's a real
hurdle for me, especially in contexts/languages where I still have the
feeling of not even having reached preschool level.

The ML quickly gave a reaction and a hint how to patch it. What's
really needed would be an option to do that at runtime and especially
per socket/connection, but it is a start and shows what should be
touched.

It's an issue now: <https://github.com/gambit/gambit/issues/904>
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

yeti <yeti@tilde.institute> writes:

[...]

What I do not want is switching to Lisp. I'd prefer to only use one
Lisp and this shall be Emacs-Lisp and that one is really bad at network access and even worse for making stand alone tools to run without Emacs.

I was introduced to Lisp last December and was coming from Racket. I
had not looked back. Lisp seems so much easier and more fun to use.

Similarly for Scheme. My experiences with Debian and Ubuntu told me
long ago that very similar, but not identical is a bigger problem than
big differences, e.g. using two more distant Unix relatives side by
side.

I believe that might be true.
--- Synchronet 3.19b-Linux NewsLink 1.113

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

[...]

What I do not want is switching to Lisp. I'd prefer to only use one
Lisp and this shall be Emacs-Lisp and that one is really bad at network
access and even worse for making stand alone tools to run without Emacs.

I was introduced to Lisp last December and was coming from Racket. I
had not looked back. Lisp seems so much easier and more fun to use.

Similarly for Scheme. My experiences with Debian and Ubuntu told me
long ago that very similar, but not identical is a bigger problem than
big differences, e.g. using two more distant Unix relatives side by
side.

I believe that might be true.

|\/| _ _ _ _ _ | | |
| |(_)(_)(_)| | | . . .

I'm hallucinating and still see Patricia here...

;-)



I've no aversion against Lisp, just Emacs-Lisp makes the most sense for
me now. I'm still in Lisp and Scheme preschool, but it is fun to see
how useful even a few lines are.

E.g. in that experiment where I collect code snippets that all shall
produce the same output and suppress the output unless it does not match
the expected results.

<http://yeti.freeshell.org/brain/unsorted.d/compare-results.org> \\ <http://yeti.freeshell.org/brain/unsorted.d/compare-results.html>

I'll need many more of such small helpers and that is the reason why
Elisp makes the most sense for me. Maybe when I start to dream in Lisp
I may dare to touch a 2nd dialect.
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

yeti <yeti@tilde.institute> writes:

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

[...]

What I do not want is switching to Lisp. I'd prefer to only use one
Lisp and this shall be Emacs-Lisp and that one is really bad at network
access and even worse for making stand alone tools to run without Emacs.

I was introduced to Lisp last December and was coming from Racket. I
had not looked back. Lisp seems so much easier and more fun to use.

Similarly for Scheme. My experiences with Debian and Ubuntu told me
long ago that very similar, but not identical is a bigger problem than
big differences, e.g. using two more distant Unix relatives side by
side.

I believe that might be true.

|\/| _ _ _ _ _ | | |
| |(_)(_)(_)| | | . . .

I'm hallucinating and still see Patricia here...

;-)

LOL! That was not a lie---but I changed my mind and killed all groups
and left this one. What was annoying me was that I had subscribed to a
lot of groups and they're all quiet, so I wanted to just go away. But
I'll keep this one just to chat with you.

I've no aversion against Lisp, just Emacs-Lisp makes the most sense for
me now. I'm still in Lisp and Scheme preschool, but it is fun to see
how useful even a few lines are.

Dude, you gotta use a Lisp that doesn't require you to run emacs to
compute it. I wonder why the EMACS people didn't release an interpreter
or compiler outside of the GNU EMACS. I think that would be healthy to
the GNU EMACS community.

E.g. in that experiment where I collect code snippets that all shall
produce the same output and suppress the output unless it does not match
the expected results.

<http://yeti.freeshell.org/brain/unsorted.d/compare-results.org> \\ <http://yeti.freeshell.org/brain/unsorted.d/compare-results.html>

Hey, you're interested in numbers. Cool. I just proved the fundamental theorem of arithmetic (after proving Euclid's Lemma).

I'll need many more of such small helpers and that is the reason why
Elisp makes the most sense for me. Maybe when I start to dream in Lisp
I may dare to touch a 2nd dialect.

Lisp is great and it's not too far away from Elisp. (I loved it.)
--- Synchronet 3.19b-Linux NewsLink 1.113

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

|\/| _ _ _ _ _ | | |
| |(_)(_)(_)| | | . . .

I'm hallucinating and still see Patricia here...

;-)

LOL! That was not a lie---but I changed my mind and killed all groups
and left this one. What was annoying me was that I had subscribed to a
lot of groups and they're all quiet, so I wanted to just go away. But
I'll keep this one just to chat with you.

Oh ... nice ... but being one of many here would even be nicer.

Dude, you gotta use a Lisp that doesn't require you to run emacs to
compute it. I wonder why the EMACS people didn't release an interpreter
or compiler outside of the GNU EMACS. I think that would be healthy to
the GNU EMACS community.

Emacs can be started as s script interpreter without staring all the UI
parts, but that doesn't change it's capabilities and I'd perfer a
"smaller" Lisp that compiles to C and/or has a good FFI. Systems that
ignore that layer feel like a wart on a Unix. I see Elisp only in the
context of extending Emacs and especially nice helpers in Org/Babel. So
I'll give other Lisps the next chance when I'm confident that they will
not confuse me when using Elisp in Org/Babel.

...aaaaand there is the Lisp used in Sawfish (Rep) which claims to be
like Elisp. I never used it outside of Sawfish, but configuring that WM
via Lisp was not a big problem even after not having touched Lisp for
decades. I played with Sawfish round about the Debian6-to-7 upgrade
era.

<http://sawfish.tuxfamily.org/> <http://sawfish.tuxfamily.org/librep.html/index.html>

E.g. in that experiment where I collect code snippets that all shall
produce the same output and suppress the output unless it does not match
the expected results.

<http://yeti.freeshell.org/brain/unsorted.d/compare-results.org> \\
<http://yeti.freeshell.org/brain/unsorted.d/compare-results.html>

Hey, you're interested in numbers. Cool. I just proved the fundamental theorem of arithmetic (after proving Euclid's Lemma).

-----------------------------------------------------------------------
3. Hitchhiker 1: (6) He stood up straight and looked the world squarely
in the fields and hills. To add weight to his words he stuck the rabbit
bone in his hair. He spread his arms out wide. (7) "I will go mad!" he announced. -----------------------------------------------------------------------

Primes drive me mad. In a positive way.

I had an overdose of university maths and forgot most of it. I still occasionally like to do weird things with and without maths. Sometimes
it is just comparing code snippets in multiple languages like an own
little Rosetta Code, sometimes numbers or maths are more in the
spotlight than the languages.

Short form: I loved maths until the university destroyed that and now
I only play a bit with it.

The last fun I hat was "transposing" a prime number algorithm I made
years ago by "pulling the sieve of Eratosthenes through a Klein's
bottle" to using unary numbers in plain SH. If run open ended normally
the original sieve would need to nest infinite loops, and I just
reordered that to a more sequential way. That needs storage and
original SH lacks arrays, but you can fake them by constructing variable
names like S$SOMENUMBER. Combined with an unary number representation
that then was kind of funny.

Best Klein's bottle clip:

<https://www.youtube.com/watch?v=cdLXvJ8wFY4>

So basically I translated this Python2 snippet to SH without BASH or ZSH blingbling:

------------------------------------------------------------------------ #!/usr/bin/env python
#
# floating (sliding?) sieve of eratosthenes prime number generator #==================================================================
#
# 2011-09-12, yeti
#

L = {}
n = 2

while 1:

if n in L:
P = L[n]
del L[n] # optional - saves some memory.
else:
P = n
print n

m = n+P
while m in L:
m += P
L[m] = P

n += 1
----------------------------------------------------------

This way all action happens at a hotspot of the number line plus
maximally a distance in which the already known primes jump around like
frogs on a one dimensional playing field and they jump by the distance
of their own value. Or boring: Just another serialisation of the
original sieve. Getting it done in SH was the hurdle because I never
used SH's eval function before. Somehow KSH, BASH and other shells had
all the blingbling that made fighting with eval unnecessary, lile
${!name} for indirect references.

And than unary did bite me again, I turned the 1st mess I used in the
primes example into some functions, added whole numbers as pairs of
credit and debit (C.D) to avoid an explicit zero and then fractions as
pairs of a whole number as numerator N and a natural number as scale S
(wanted to avoid denumerator as name because I already used D for debit)
N:S = C.D:S.

Guess what... I added enough functions to theoretically calculate a
Mandelbrot set in unary fractions, but it turned out that the size of
the unary number's representations would blow up every swap size I can
provide.

Ok. Hit shappens! But it was worth it.

Looking at the whole numbers C.D and the operation to reduce them to a representation where the bigger part is reduced by the amount of the
other one (subtract the "greatest common subtractor") that was so
similar to trimming down quotients by dividing away the GCD, that I
smiled for days over that analogy.

That alone was worth it and compensated well for not getting a purely
unary calculated Mandelbrot set image.

Lisp is great and it's not too far away from Elisp. (I loved it.)

Maybe I'll get there later.
--
I'm nonbinary, I use Lisp and Scheme. -- 20240301/yeti
--- Synchronet 3.19b-Linux NewsLink 1.113

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

|\/| _ _ _ _ _ | | |
| |(_)(_)(_)| | | . . .

I'm hallucinating and still see Patricia here...

;-)

LOL! That was not a lie---but I changed my mind and killed all groups
and left this one. What was annoying me was that I had subscribed to a
lot of groups and they're all quiet, so I wanted to just go away. But
I'll keep this one just to chat with you.

Oh ... nice ... but being one of many here would even be nicer.

Dude, you gotta use a Lisp that doesn't require you to run emacs to
compute it. I wonder why the EMACS people didn't release an interpreter
or compiler outside of the GNU EMACS. I think that would be healthy to
the GNU EMACS community.

Emacs can be started as s script interpreter without staring all the UI
parts, but that doesn't change it's capabilities and I'd perfer a
"smaller" Lisp that compiles to C and/or has a good FFI. Systems that
ignore that layer feel like a wart on a Unix. I see Elisp only in the
context of extending Emacs and especially nice helpers in Org/Babel. So
I'll give other Lisps the next chance when I'm confident that they will
not confuse me when using Elisp in Org/Babel.

...aaaaand there is the Lisp used in Sawfish (Rep) which claims to be
like Elisp. I never used it outside of Sawfish, but configuring that WM
via Lisp was not a big problem even after not having touched Lisp for
decades. I played with Sawfish round about the Debian6-to-7 upgrade
era.

<http://sawfish.tuxfamily.org/> <http://sawfish.tuxfamily.org/librep.html/index.html>

E.g. in that experiment where I collect code snippets that all shall
produce the same output and suppress the output unless it does not match
the expected results.

<http://yeti.freeshell.org/brain/unsorted.d/compare-results.org> \\
<http://yeti.freeshell.org/brain/unsorted.d/compare-results.html>

Hey, you're interested in numbers. Cool. I just proved the fundamental theorem of arithmetic (after proving Euclid's Lemma).

-----------------------------------------------------------------------
3. Hitchhiker 1: (6) He stood up straight and looked the world squarely
in the fields and hills. To add weight to his words he stuck the rabbit
bone in his hair. He spread his arms out wide. (7) "I will go mad!" he announced. -----------------------------------------------------------------------

Primes drive me mad. In a positive way.

I had an overdose of university maths and forgot most of it. I still occasionally like to do weird things with and without maths. Sometimes
it is just comparing code snippets in multiple languages like an own
little Rosetta Code, sometimes numbers or maths are more in the
spotlight than the languages.

Short form: I loved maths until the university destroyed that and now
I only play a bit with it.

The last fun I hat was "transposing" a prime number algorithm I made
years ago by "pulling the sieve of Eratosthenes through a Klein's
bottle" to using unary numbers in plain SH. If run open ended normally
the original sieve would need to nest infinite loops, and I just
reordered that to a more sequential way. That needs storage and
original SH lacks arrays, but you can fake them by constructing variable
names like S$SOMENUMBER. Combined with an unary number representation
that then was kind of funny.

Best Klein's bottle clip:

<https://www.youtube.com/watch?v=cdLXvJ8wFY4>

So basically I translated this Python2 snippet to SH without BASH or ZSH blingbling:

------------------------------------------------------------------------ #!/usr/bin/env python
#
# floating (sliding?) sieve of eratosthenes prime number generator #==================================================================
#
# 2011-09-12, yeti
#

L = {}
n = 2

while 1:

if n in L:
P = L[n]
del L[n] # optional - saves some memory.
else:
P = n
print n

m = n+P
while m in L:
m += P
L[m] = P

n += 1
----------------------------------------------------------

This way all action happens at a hotspot of the number line plus
maximally a distance in which the already known primes jump around like
frogs on a one dimensional playing field and they jump by the distance
of their own value. Or boring: Just another serialisation of the
original sieve. Getting it done in SH was the hurdle because I never
used SH's eval function before. Somehow KSH, BASH and other shells had
all the blingbling that made fighting with eval unnecessary, lile
${!name} for indirect references.

And than unary did bite me again, I turned the 1st mess I used in the
primes example into some functions, added whole numbers as pairs of
credit and debit (C.D) to avoid an explicit sign and then fractions as
pairs of a whole number as numerator N and a natural number as scale S
(wanted to avoid denumerator as name because I already used D for debit)
N:S = C.D:S.

Guess what... I added enough functions to theoretically calculate a
Mandelbrot set in unary fractions, but it turned out that the size of
the unary number's representations would blow up every swap size I can
provide.

Ok. Hit shappens! But it was worth it.

Looking at the whole numbers C.D and the operation to reduce them to a representation where the bigger part is reduced by the amount of the
other one (subtract the "greatest common subtractor") that was so
similar to trimming down quotients by dividing away the GCD, that I
smiled for days over that analogy.

That alone was worth it and compensated well for not getting a purely
unary calculated Mandelbrot set image.

Lisp is great and it's not too far away from Elisp. (I loved it.)

Maybe I'll get there later.
--
I'm nonbinary, I use Lisp and Scheme. -- 20240301/yeti
--- Synchronet 3.19b-Linux NewsLink 1.113

yeti <yeti@tilde.institute> writes:

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

|\/| _ _ _ _ _ | | |
| |(_)(_)(_)| | | . . .

I'm hallucinating and still see Patricia here...

;-)

LOL! That was not a lie---but I changed my mind and killed all groups
and left this one. What was annoying me was that I had subscribed to a
lot of groups and they're all quiet, so I wanted to just go away. But
I'll keep this one just to chat with you.

Oh ... nice ... but being one of many here would even be nicer.

Dude, you gotta use a Lisp that doesn't require you to run emacs to
compute it. I wonder why the EMACS people didn't release an interpreter
or compiler outside of the GNU EMACS. I think that would be healthy to
the GNU EMACS community.

Emacs can be started as s script interpreter without staring all the UI parts,

Does it also have a REPL? :-) I know you'll say the GNU EMACS, but I
don't think the GNU EMACS offers a typical (ELISP) REPL.

but that doesn't change it's capabilities and I'd perfer a "smaller"
Lisp that compiles to C and/or has a good FFI.

Most have FFI, but not all of them to compile to C. I don't know what's
a good FFI. Lisp has cffi, for example, which seems nice enough.

...aaaaand there is the Lisp used in Sawfish (Rep) which claims to be
like Elisp. I never used it outside of Sawfish, but configuring that WM
via Lisp was not a big problem even after not having touched Lisp for decades. I played with Sawfish round about the Debian6-to-7 upgrade
era.

<http://sawfish.tuxfamily.org/> <http://sawfish.tuxfamily.org/librep.html/index.html>

Nice! Thanks for sharing.

Hey, you're interested in numbers. Cool. I just proved the fundamental
theorem of arithmetic (after proving Euclid's Lemma).

-----------------------------------------------------------------------
3. Hitchhiker 1: (6) He stood up straight and looked the world squarely
in the fields and hills. To add weight to his words he stuck the rabbit
bone in his hair. He spread his arms out wide. (7) "I will go mad!" he announced. -----------------------------------------------------------------------

Primes drive me mad. In a positive way.

I had an overdose of university maths and forgot most of it. I still occasionally like to do weird things with and without maths. Sometimes
it is just comparing code snippets in multiple languages like an own
little Rosetta Code, sometimes numbers or maths are more in the
spotlight than the languages.

Short form: I loved maths until the university destroyed that and now
I only play a bit with it.

That's sad. What major you were in? At the university I realized my
education wasn't good but I had a solid understanding of computers and I
asked why did that happen---the answer was ``school never taught me
anything about computers''. What did I do then? I just stopped paying attention to classes and designed my own curriculum. (And I decided not
to major in computer science because that would have killed my passion
for the subject.) Grades went down to a minimum, but I graduated
satisfied and have been harvesting the profits ever since.

It's a fraud. A university is there to certify you, not to help you in
any meaningful way. Surely you can find people to teach and learn with,
but that's not the university---that's the people.

The last fun I hat was "transposing" a prime number algorithm I made
years ago by "pulling the sieve of Eratosthenes through a Klein's
bottle" to using unary numbers in plain SH.

Sounds incomprehensible. :)
--- Synchronet 3.19b-Linux NewsLink 1.113

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

Emacs can be started as s script interpreter without staring all the UI
parts,

Does it also have a REPL? :-) I know you'll say the GNU EMACS, but I
don't think the GNU EMACS offers a typical (ELISP) REPL.

I've not yet seen an Elisp REPL outside of Emacs, but it might exist.
Inside Emacs I prefer Eshell as REPL over e.g. the scratch buffer others
seem to like more. Eshell mimicks a Unix shell but is split brained
allowing Lisp and shell like notation. Some typical commands (grep,
less, ls, ...) are rebuilt via Lisp functions, external commands can be
called like in other shells. Overlaid commands like ls can be accesse
by prefixing them with * and ib most cases I prefer the external
originals even in Eshell.

...

Short form: I loved maths until the university destroyed that and now
I only play a bit with it.

That's sad. What major you were in?

I started with Maths plus IT as secondary topic, but when that took too
long because I lost all the fun and working with computers in an
institute was nicer than the fight with the university, I switched them,
so IT got my main subject. It already was my hobby and when I started
with Maths there it only could be secondary subject, but when I was fed
up with Maths, IT was promoted to be possible as main subject too.

At the university I realized my education wasn't good but I had a
solid understanding of computers and I asked why did that happen---the
answer was ``school never taught me anything about computers''. What
did I do then?

I scared my maths teachers at school and they were happy when I just
wrote programs instead of disturbing them with questions they couldn't
answer.

At the university there were several threads of maths courses/topics in parallel and we used stuff in some of them we hadn't introduced and so
were not allowed to use in others some hours earlier or later. That
tohubohu was absolutely demotivating. Additionally one could fail in
courses just for not remembering names despite deducing and applying all formulae and logic correctly. I hadn't planned to learn history, I had expected to learn and practice Maths and there the formulae are the
condensed truths.
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

Patricia Ferreira <pferreira@example.com> writes:

yeti <yeti@tilde.institute> writes:

Emacs can be started as s script interpreter without staring all the UI
parts,

Does it also have a REPL? :-) I know you'll say the GNU EMACS, but I
don't think the GNU EMACS offers a typical (ELISP) REPL.

I've not yet seen an Elisp REPL outside of Emacs, but it might exist.
Inside Emacs I prefer Eshell as REPL over e.g. the scratch buffer others
seem to like more. Eshell mimicks a Unix shell but is split brained
allowing Lisp and shell like notation. Some typical commands (grep,
less, ls, ...) are rebuilt via Lisp functions, external commands can be
called like in other shells. Overlaid commands like ls can be accessed
by prefixing them with * and in most cases I prefer the external
originals even in Eshell.

...

Short form: I loved maths until the university destroyed that and now
I only play a bit with it.

That's sad. What major you were in?

I started with Maths plus IT as secondary topic, but when that took too
long because I lost all the fun and working with computers in an
institute was nicer than the fight with the university, I switched them,
so IT got my main subject. It already was my hobby and when I started
with Maths there it only could be secondary subject, but when I was fed
up with Maths, IT was promoted to be possible as main subject too.

At the university I realized my education wasn't good but I had a
solid understanding of computers and I asked why did that happen---the
answer was ``school never taught me anything about computers''. What
did I do then?

I scared my maths teachers at school and they were happy when I just
wrote programs instead of disturbing them with questions they couldn't
answer.

At the university there were several threads of maths courses/topics in parallel and we used stuff in some of them we hadn't introduced and so
were not allowed to use in others some hours earlier or later. That
tohubohu was absolutely demotivating. Additionally one could fail in
courses just for not remembering names despite deducing and applying all formulae and logic correctly. I hadn't planned to learn history, I had expected to learn and practice Maths and there the formulae are the
condensed truths.
--
I do not bite, I just want to play.
--- Synchronet 3.19b-Linux NewsLink 1.113

And another mystery was quickly answered[0] within minutes and I could
not even have been faster if I had not asked the ML first. Well, at
least they aren't yelling at me for this order of interaction.

This time it really looked strange. Code worked in the interpreter, but
not in compiled form, depending on having an `(import ...)` line or not.

Executable does NOTHING, when importing srfi-module (but code works

in REPL) #872 \\
https://github.com/gambit/gambit/issues/872

I declare not looking up the issues first as mild form of PEBCAK. Ok?
But with new stuff and not feeling at home in their terminology and
jargon often does not make finding stuff easier. So I ask for applying extenuating circumstances.

____________

[0]: The previous one got explained, so at least I can drop the PEBCAK
hypothesis, but still needs an implementation in a post build
Gambit time way.
--
3. Hitchhiker 10: (81) The Hitch Hiker's Guide to the Galaxy has this to
say on the subject of flying. (82) There is an art, it says, or rather
a knack to flying. (83) The knack lies in learning how to throw
yourself at the ground and miss.
--- Synchronet 3.19b-Linux NewsLink 1.113