💾 Archived View for gemi.dev › gemini-mailing-list › 000236.gmi captured on 2024-08-18 at 23:55:05. Gemini links have been rewritten to link to archived content

View Raw

More Information

⬅️ Previous capture (2023-12-28)

-=-=-=-=-=-=-

[ANN] Tiny ad-hoc server

1. James Tomasino (tomasino (a) lavabit.com)

I put up a tiny experiment of an ad-hoc server over on 
gemini://tomasino.org. The code for the server is on that gemini page for 
you to inspect and play with. Credit to xq for the concept. It's trivial 
to implement and may open the door for some folks who don't have a plan of 
running a big server. It may even be able to work with solderpunk's 
micro-instance ideas. I'm not sure if there's a way to validate client 
certs by building on to it, but that would make for easy micro-services.

- tomasino

Link to individual message.

2. Blair Vidakovich (vidak (a) riseup.net)

wow... that is awesome, i can't believe it is done in so few lines of
code!

vidak

-eof

Link to individual message.

3. defdefred (defdefred (a) protonmail.com)

Hello!

I don't understand how the infinite while works in this case...
Is it consuming cpu all the time?

I play once with ncat and the program was rather in two parts:

1: the-worker
#!/bin/sh
printf "20 text/gemini; charset=utf-8\r\n"
printf "Page generated: %s\n\n" "$(date)"
cat file-to-serve

2: the-listener
#!/bin/sh

port="1965"
cert="cert.pem"
key="key.pem"

printf "Starting gemini server on port %s...\n" "$port"
ncat --ssl --ssl-cert "$cert" --ssl-key "$key" --listen "$port" -k -c ./the-worker

freD.


??????? Original Message ???????
On Friday 19 June 2020 19:28, James Tomasino <tomasino at lavabit.com> wrote:

&gt; I put up a tiny experiment of an ad-hoc server over on 
gemini://tomasino.org. The code for the server is on that gemini page for 
you to inspect and play with. Credit to xq for the concept. It's trivial 
to implement and may open the door for some folks who don't have a plan of 
running a big server. It may even be able to work with solderpunk's 
micro-instance ideas. I'm not sure if there's a way to validate client 
certs by building on to it, but that would make for easy micro-services.
&gt;
&gt; -   tomasino

</tomasino at lavabit.com>

Link to individual message.

4. James Tomasino (tomasino (a) lavabit.com)

On 6/25/20 11:34 AM, defdefred wrote:
> I don't understand how the infinite while works in this case...
> Is it consuming cpu all the time?

ncat waits only for a single connection, then terminates. I tried using 
the switch to keep the connection alive and reuse it, but that means the 
connection is never closed on any single request and most clients display 
nothing. By wrapping it in an infinite while I just tell ncat to start a 
new listener when the previous one is used. There's only ever one running 
at a time, and it's a single thread, but traffic is low and it seems to work just fine.

Link to individual message.

5. defdefred (defdefred (a) protonmail.com)

Ok I'm dumb... don't saw the braces :-)

Bash process substitution will do too

#!/bin/sh

port="1965"
cert="cert.pem"
key="key.pem"

printf "Starting gemini server on port %s...\n" "$port"
while true; do
    ncat --ssl --ssl-cert "$cert" --ssl-key "$key" --listen "$port" <( \
    printf "20 text/gemini; charset=utf-8\r\n" ; \
    printf "Page generated: %s\n\n" "$(date)" ; \
    cat "$1" )
done

freD.

??????? Original Message ???????
On Thursday 25 June 2020 14:09, James Tomasino <tomasino at lavabit.com> wrote:

> On 6/25/20 11:34 AM, defdefred wrote:
>
> > I don't understand how the infinite while works in this case...
> > Is it consuming cpu all the time?
>
> ncat waits only for a single connection, then terminates. I tried using 
the switch to keep the connection alive and reuse it, but that means the 
connection is never closed on any single request and most clients display 
nothing. By wrapping it in an infinite while I just tell ncat to start a 
new listener when the previous one is used. There's only ever one running 
at a time, and it's a single thread, but traffic is low and it seems to work just fine.

Link to individual message.

6. Case Nelson (me (a) case.codes)

I think you can get rid of the loop here and use ncat -k , it's been 
working pretty well for me in testing.

Jun 25, 2020 6:30:42 AM defdefred <defdefred at protonmail.com>:

> Ok I'm dumb... don't saw the braces :-)
> 
> Bash process substitution will do too
> 
> #!/bin/sh
> 
> port="1965"
> cert="cert.pem"
> key="key.pem"
> 
> printf "Starting gemini server on port %s...\n" "$port"
> while true; do
> ncat --ssl --ssl-cert "$cert" --ssl-key "$key" --listen "$port" <( \
> printf "20 text/gemini; charset=utf-8\r\n" ; \
> printf "Page generated: %s\n\n" "$(date)" ; \
> cat "$1" )
> done
> 
> freD.
> 
> ??????? Original Message ???????
> On Thursday 25 June 2020 14:09, James Tomasino <tomasino at lavabit.com> wrote:
> 
>> On 6/25/20 11:34 AM, defdefred wrote:
>> 
>>> I don't understand how the infinite while works in this case...
>>> Is it consuming cpu all the time?
>> 
>> ncat waits only for a single connection, then terminates. I tried using 
the switch to keep the connection alive and reuse it, but that means the 
connection is never closed on any single request and most clients display 
nothing. By wrapping it in an infinite while I just tell ncat to start a 
new listener when the previous one is used. There's only ever one running 
at a time, and it's a single thread, but traffic is low and it seems to work just fine.
>

Link to individual message.

---

Previous Thread: Great web site to port to gemini

Next Thread: Some new tests in the Gemini Client Torture Test