Hostname is not the issue, it can connect and then fails some handshake or whatever. Adding verbosity just seems to hexdump the packets in addition to logging an error.
[2025.11.26-09:10:21.460] [FINE ] got server SSH_MSG_KEX_ECDH_INIT
sending: length = 48
0000 00 00 00 2C 06 1E 00 00 00 20 D0 91 2C 8F 57 AF ...,..... ..,.W.
> Short: BebboSSH – SSH2 suite (client/server, sftp) with modern ciphers
Sweet.
> It will work on an unaccelerated Amiga but establishing the connection takes
about one minute.
Is latency good once the connection is established? Speed tests in the kB/s seem promising, but for interactive use that seems like the determining factor.
Also, slightly meta: What is this web ... app(?)? Looks like a git frontend?
At 40kbps I don't think the latency can be great. Having done SSH over 64kbps when I overran my mobile data subscription, my experience is that modern SSH clients expect more than that to run smoothly.
> Also, slightly meta: What is this web ... app(?)? Looks like a git frontend?
Looking at the source, this seems like a custom-built git frontend, served by a bespoke web server called BEJY (by the same author) it seems.
> At 40kbps I don't think the latency can be great. Having done SSH over 64kbps when I overran my mobile data subscription, my experience is that modern SSH clients expect more than that to run smoothly.
What do these bandwidth numbers have to do with latency?
Even with good latency, most of my connection was taken up by TCP retransmissions at that bandwidth because of an over-eager (standard) SSH client. I can't imagine the TCP stack on an Amiga doing much better in the same scenario.
Not exactly the same scenario of course, but I'd expect you'd need a custom SSH client willing to wait with small enough packets to keep the latency acceptable.
If the round trip time is dominated by the time taken to encrypt and decrypt packets locally, as seems here, then the speed at which it can complete that is absolutely important for measuring "latency".
I gave it a go (A1200+AmiTCP 4.3), doesn't seem to work unfortunately and I can't seem to find any issue tracker on the website.
3.RAM Disk:bebbossh> bebbossh user@10.0.0.1
[2025.11.26-09:10:01.280] [INFO ] can't open `envarc:.ssh/ssh_config`
[2025.11.26-09:10:01.640] [ERROR] can't read 4 uint8_t header, got 0
[2025.11.26-09:10:01.647] [ERROR] can't read SSH_MSG_KEX_ECDH_REPLY
ERROR: 16 - tcp read failed
The program seems to call gethostbyname() on the supplied host, so you may want to try providing a DNS name instead.
Also, call with -v8 to get more detailed logging.
Hostname is not the issue, it can connect and then fails some handshake or whatever. Adding verbosity just seems to hexdump the packets in addition to logging an error.
[2025.11.26-09:10:21.460] [FINE ] got server SSH_MSG_KEX_ECDH_INIT
sending: length = 48
0000 00 00 00 2C 06 1E 00 00 00 20 D0 91 2C 8F 57 AF ...,..... ..,.W.
0010 91 B2 F9 C5 77 42 30 D1 2E A5 A0 B2 C6 C8 76 4C ....wB0.......vL
0020 5F 32 27 E7 1F 0D A2 32 C0 3A 00 00 00 00 00 00 _2'....2.:......
[2025.11.26-09:10:21.520] [FINE ] sent server SSH_MSG_KEX_ECDH_INIT
[2025.11.26-09:10:21.533] [ERROR] can't read 4 uint8_t header, got 0
[2025.11.26-09:10:21.540] [ERROR] can't read SSH_MSG_KEX_ECDH_REPLY
[2025.11.26-09:10:21.554] [FINE ] bye bye
ERROR: 16 - tcp read failed
Well, what's in envarc:.ssh/ssh_config ? :-)
> Short: BebboSSH – SSH2 suite (client/server, sftp) with modern ciphers
Sweet.
> It will work on an unaccelerated Amiga but establishing the connection takes about one minute.
Is latency good once the connection is established? Speed tests in the kB/s seem promising, but for interactive use that seems like the determining factor.
Also, slightly meta: What is this web ... app(?)? Looks like a git frontend?
At 40kbps I don't think the latency can be great. Having done SSH over 64kbps when I overran my mobile data subscription, my experience is that modern SSH clients expect more than that to run smoothly.
> Also, slightly meta: What is this web ... app(?)? Looks like a git frontend?
Looking at the source, this seems like a custom-built git frontend, served by a bespoke web server called BEJY (by the same author) it seems.
> At 40kbps I don't think the latency can be great. Having done SSH over 64kbps when I overran my mobile data subscription, my experience is that modern SSH clients expect more than that to run smoothly.
What do these bandwidth numbers have to do with latency?
Even with good latency, most of my connection was taken up by TCP retransmissions at that bandwidth because of an over-eager (standard) SSH client. I can't imagine the TCP stack on an Amiga doing much better in the same scenario.
Not exactly the same scenario of course, but I'd expect you'd need a custom SSH client willing to wait with small enough packets to keep the latency acceptable.
Interesting! Thanks. I had no idea the standard SSH client was so eager.
If the round trip time is dominated by the time taken to encrypt and decrypt packets locally, as seems here, then the speed at which it can complete that is absolutely important for measuring "latency".
Sure. But if it can decrypt at 40 kbps, then this won't be the bottleneck.
> Also, slightly meta: What is this web ... app(?)? Looks like a git frontend?
Seems to be a custom git frontend written by him: https://franke.ms/git/bebbo/tigler
[dead]
https://github.com/eisbaw/nano_ssh_server
“World's smallest self-contained SSH server. Entirely AI slop”
:-|
I wonder why the name change? Threats from rights holders?
I wouldn't be surprised if this was the case, given the long history[0].
0. https://sites.google.com/site/amigadocuments/