1. 16

Screenshots: http://eab.abime.net/showthread.php?t=108120

  1.  

  2. 1

    Requirements are Workbench 3.0, a TCP/IP stack and AmiSSL. AmiSSL requires at least a 68020.

    It’s nice to see Gemini support on retrocomputers. It’s a frequent topic of discussion on the Gemini mailing list as to what extent the TLS requirement rules out older computers. The consensus seems to be that 8-bit micros will need some kind of coprocessor, but that 80386 or 68020 systems should be fine. It’s good to have a proof-by-example.

    1. 2

      I wonder how well it works with a 68020. AmiSSL is an OpenSSL port. I’ve never tried it, but some time back I was looking at mBedTLS (a more efficient TLS stack that targets embedded things but has 68k support). The folks who ported mBed to mac 68k said that anything slower than a 68040 took so long on the handshake that most servers time out.:

      From my limited testing so far, a 33mhz 68040 processor will complete a TLS handshake in about 15 seconds. Whereas a 16mhz 68030 will take about 70 seconds - during which time the remote server will timeout. So it looks like the minimum requirement is a 68040.

      So I’m not sure this constitutes a proof by example till you try it on an ’020.

      1. 1

        Huh, interesting. I feel like a 16-mhz 68030 is a pretty cromulent processor — it’s what the Mac SE/30 had, and that’s kind of my benchmark for “1990s personal Unix machine”.

        1. 2

          I agree. I drove a PowerBook 180 (which basically performed like an SE/30… 33MHz in a PowerBook was not as much faster as the numbers there suggest) that I rescued off the scrap heap for quite a long time. But I’m sure I didn’t use even an early version of TLS or SSL. That was a very nice machine to use. MPW was even OK on it.