Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Ask HN: Oldest piece of hardware a user could access the internet with?

news.ycombinator.com

Post reply on HN
Print view View original

41–50 of 54 posts

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#41

Post by bawolff »

classichasclass wrote: post #20
throwitaway1235 wrote: post #6

I just realized the semi-stupidity of my question. I should have said "modern web", considering the ARPANET could be considered a type of "internet" (?) that ran using hardware from 1969? Thank you for the replies though, something to read up on.

Do you mean something that, for example, could access TLS 1.2? I hacked up such a thing that works on a number of my old boxes, including NeXSTEP and A/UX. The oldest system I got it to work on is a Mac IIci running NetBSD without a cache card; it needs about 22 seconds to do a TLS 1.2 transaction with a local server, but it works. https://oldvcr.blogspot.com/2020/11/fun-with-crypto-ancienne...

Huh, wonder where its spending all that time and if different ciphers would make a significant difference. I imagine many of the modern ciphers have hand tuned assembly for 64bit computers but probably not for older computers.

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#42

Post by einr »

msla wrote: post #28
h2odragon wrote: post #7

"access"... If you wanted, an old vt100/vt220 or whatever other hardware serial terminal you like can be connected to a linux or other system and access a shell account with text browsers etc. In that scenario the terminal is "accessing the internet". (Did this with a Televideo 970 in 2007 iirc for the ebay listing where we sold it) If you want something to run an IP stack, then a 286 running `ka9qnos` is about as sm…

> "access"... If you wanted, an old vt100/vt220 or whatever other hardware serial terminal you like can be connected to a linux or other system and access a shell account with text browsers etc. In that scenario the terminal is "accessing the internet". (Did this with a Televideo 970 in 2007 iirc for the ebay listing where we sold it) By this standard, I'm accessing the Internet using a flat-panel display. If somethi…

> > There might be a c64 IP stack but at that point its getting technicality again; how much access and function counts?

> Being able to talk at least IP and TCP without external help.

Contiki (https://www.c64-wiki.com/wiki/Contiki) runs on a stock C64 (plus an Ethernet adapter) and implements a full TCP/IP stack. It has IRC, FTP, e-mail, a web browser, etc. No tricks -- it runs on the hardware with no external help.

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#45

Post by throw0101a »

tbenst wrote: post #9
johndoe0815 wrote: post #3

"Accessing the internet" is a fairly broad term. In general, all machines supporting IPv4 and some sort of interface (UART for SLIP/PPP or Ethernet) can access the internet. However, older machines will not be able to use more modern services like ssh or the web due to compute power or memory limitations and you will obviously not be able to connect to IPv6-only hosts. So, for example a PDP11 running 2.9BSD (e.g. a 1…

Thank you for the history lesson! I had no idea IPv4 was so old

* https://www.simonandschuster.com/books/Where-Wizards-Stay-Up...

* https://www.goodreads.com/book/show/281818.Where_Wizards_Sta...

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#46

Post by fogihujy »

_b8r0 wrote: post #27
throwitaway1235 wrote: post #6

I just realized the semi-stupidity of my question. I should have said "modern web", considering the ARPANET could be considered a type of "internet" (?) that ran using hardware from 1969? Thank you for the replies though, something to read up on.

I regularly use the Web most days on an Amiga 4000, built in 1992. It's not perfect for everything, but I can use things like search engines, visit web sites over HTTPS etc. I mostly visit Amiga-specific sites that tend to have lower overheads. I've also visited sites like like Twitter, Facebook, CNN etc. in the past. It should be possible to use older machines to browse the web but it stops being about the machine a…

Is the 68k port of Netsurf still kept up to date? I assume it would be slow, but Netsurf itself is quite usable, though I wouldn't say it's a modern Internet experience.

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#47

Post by cestith »

What amount of assistance are we talking? There are IP stacks for IBM PC with DOS and for C64. I have Ethernet for my 5150. I have a device that turns a Raspberry Pi into a peripheral for a Ti-99/4A which can provide an Internet connection to the TI machine. I have another device that uses an ESP2866 or so to get anything with a serial port onto wifi. I a DECMate III which is a PDP-8 class machine that can get on the web. I have a couple of Livingston Portmaster 2e access concentrators that can give dozens of systems IP access via SLIP, PPP, or bare terminal-to-IP access. My 286 with OS/2 works fine for graphical browsing, but anything approaching modern TLS ciphers are a performance nightmare.

At a minimum, you're looking at as old as the 68020 Macs and the 386 PCs. The PDP-10, PDP-11, Amiga 1000, the earliest Sun machines (68k, pre-Sparc), and similar age machines from the 1980s and early 1990s having a really reasonable amount of functionality with the right options, OS version, drivers, and software.

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#48

Post by _b8r0 »

fogihujy wrote: post #46
_b8r0 wrote: post #27

Earlier quoted context omitted.

I regularly use the Web most days on an Amiga 4000, built in 1992. It's not perfect for everything, but I can use things like search engines, visit web sites over HTTPS etc. I mostly visit Amiga-specific sites that tend to have lower overheads. I've also visited sites like like Twitter, Facebook, CNN etc. in the past. It should be possible to use older machines to browse the web but it stops being about the machine a…

Is the 68k port of Netsurf still kept up to date? I assume it would be slow, but Netsurf itself is quite usable, though I wouldn't say it's a modern Internet experience.

Not really. I'm not sure where I got it from but I have several Netsurf versions on my Amiga, including a crash-prone 3.10 build. The last official release is 3.6 I think.

Even so the 3.10 build I have isn't built against current AmiSSL so TLS sites fail. It is slower than say, Voyager Browser and much slower than IBrowse. Netsurf 3.6 is usable if you're patient on an 060/50. Performance is around what I'd expect from a Pentium 75-100 maybe?

Re: Ask HN: Oldest piece of hardware a user could access the internet with?

#49

Post by Hizonner »

johndoe0815 wrote: post #3

"Accessing the internet" is a fairly broad term. In general, all machines supporting IPv4 and some sort of interface (UART for SLIP/PPP or Ethernet) can access the internet. However, older machines will not be able to use more modern services like ssh or the web due to compute power or memory limitations and you will obviously not be able to connect to IPv6-only hosts. So, for example a PDP11 running 2.9BSD (e.g. a 1…

OK, but wait. 2.9BSD is from like 1982. And I think SLIP was later than that, and PPP was definitely much later than that. So you don't have a "pure" 1975 machine; you're using later software. What does "support" mean, then?

If you're allowed to use software that's later than your hardware, then you can write a stack for almost anything, so the idea of a machine "supporting IPv4" becomes kind of meaningless. You need 576 bytes of memory for a packet buffer, plus your program. Any mid-sixties mainframe, maybe even older.

If you do count the age of the software in deciding what's "supported", and you require that the TCP/IP stack run locally and communicate with modern computers without translation, then you'll have a hard limit at about 1980 or 1981, since the TCP and IPv4 protocols weren't defined until then, and there was therefore definitely no software.

You could allow using NCP as a "legitimate Internet precursor" to TCP/IP, but if you do that there's no real reason not to just allow serial terminals as "legitimate precursors" to NCP. So 1980 or 1981 seems like the limit for a "pure" system with its own software.

If you allow software newer than the hardware, but limit yourself to not writing that software, my guess is that the oldest thing would be a KA-10, which is a 1968 computer. RFC 801 says that as of 1981, there was a stack for TENEX, which you could run on a KA-10. You could at least do Telnet, maybe FTP. That's assuming you can find another old computer to gateway whatever ancient serial protocols it was using at the physical layer.

Post reply on HN