Localtalk was great, RS-422 is a low-cost long-distance serial transport, and Apple took advantage of that to build networking into the Macintosh. Our computer lab in high school used Farallon PhoneNET transceivers, which made it easy to daisy chain computers using low-cost Category 1 RJ-11 phone wiring.
In college (Miami University), our computer lab also used PhoneNET (it really simplified wiring) combined with a Cayman Gatorbox to bridge to the campus TCP/IP backbone. By modern standards, painfully slow, but in 1992 the whole university was connected over a single 56K uplink to Columbus, so we hardly noticed.
I have fond memories of playing Spectre with networked opponents over LocalTalk.
To get online from our dorm rooms or offcampus, the only option was dial-up to the university phonebank, a 2400 baud terminal. Using the LocalTalk connection from a lab was much more responsive.
It was always some version of Phonenet (Localtalk over RJ-11 phone cabling) in my experience; I don't think I ever saw real Apple Localtalk hardware!
The Appletalk protocols were excellent for small networks. It was truly plug-and-play. From the late 1980s, if there was more than one Mac in the same building, they were probably networked together. It really was as easy as connecting a cable between them. They would auto-configure, and off you went, sharing files or playing multi-user games. This was the same era of recompiling UNIX kernels to tweak NFS buffer parameters, and when the DOS world was mired in a nightmare of TSR packet drivers and Netware servers. Very ahead of its time. I do remember the impression however, that the protocols did not scale well beyond a small LAN.
This is something worth looking into.What would a modern system built in 2023 with auto-configured everything look like?
I don't mean take Linux with some added magic - that's MacOS. I mean a full modern distributed auto-configured, plug'n'play system. Sort of a modern AS/400 meets Macintosh.
It is theoretically possible to have an entire “self configured” network today, but since a network (like a computer) that is not connected to the Internet now feels completely useless. And once you have a router it can handle a bunch of the network details via dhcp.
I do remember the impression however, that the protocols did not scale
well beyond a small LAN.
Apple deserves a lot of credit for trying to smooth over the UX. The reality wasn't that things were very much plug and pray. I absolutely remember things grinding to a halt for instance when trying to print with more than 1 or 2 computers on the network. Unfortunately networked PCs started to become fairly popular around the time that Apple went to shit. Pretty much everything > 7.5 and < 10.4 was a hot mess with the worst being MacOS 8.x and 9.x with their instability and disk corruption issues.
Trying to integrate Apple stuff into a heterogeneous network? Ughghhghgghgh. Neither AppleTalk nor MacTCP were PPC native, so with PPCs we got OpenTransport. Compared to e.g. Trumpet on Windows, I always seemed to have problems.
We ran an entire distributed multi-building school campus on AppleTalk, with multiple labs, shared printers in every building, and at least one networked staff computer in every classroom.
We absolutely handled more than two workstations using a shared printer without issue.
OpenTransport was a lot more stable than MacTCP, but that was just the TCP/IP stack. I have no idea what you mean by AppleTalk not being available on PPCs; it absolutely was, and it wasn’t replaced by OT.
It’s always the line between “plugged in and just works” and “someone is network administrating the setup” - a complex network almost always has someone smoothing over the rough edges - I remember changing the order of protocols in various pieces of equipment so that everything would work happily together.
That was definitely true for a big complicated network setup like ours, but at the same time, I ran an AppleTalk network at home with a few computers and two shared printers, and it all really did “just work”.
We connected the two different rooms with a spare twisted pair we weren’t using for the phone, and then just used daisy-chained LocalTalk PhoneNet adapters.
I seem to recall this caused one of my first exposures to Linux. I worked for a small consultancy and we set up a similar network, but with Red Hat boxes running AppleTalk to do file and print serving.
> I have fond memories of playing Spectre with networked opponents over LocalTalk.
I was in a gifted programme in primary school in the early 1990s. The teachers running it weren’t very computer-literate and expected the Macs in the classroom to only be used as a support for whatever non-computer-related intellectual activities. I got sent out of the programme back to the non-gifted classes for setting up networked Spectre and Bolo. They saw it as time-wasting gaming, but what fascinated me was that computers could be networked. I think that if I had been left to explore, I would have eventually been on the path to a computer science or IT education (and a much better salary).
I got locked out of the lan network for playing a game in high school. I don't remember which game, but I remember the teacher who ran the lab found a game stored in my directory on the fileserver. I had to beg forgiveness.
I had an Apple IIgs and a PowerMac G5 across the room from each other (still have the GS, but in the G5's spot sits a Mac Studio, now). To get files from the net onto the IIgs, I attached a Keyspan USB to serial adapter to the Mac, with the Mac-style miniDIN RS-422 connectors, and connected the GS to the Mac's serial port via RJ-11 using phoneNet adapters on each end. Worked great for filesharing to the GS. (I've got an ethernet card in the IIgs now, so I just FTP to the Mac (or wherever) to get files I want.)
Over the other side of the Atlantic, Acorn had an example of parallel evolution in the form of Econet, which used the same serial protocol, but over bigger connectors (the same 5 pin DIN that you see used in DMX networks and MIDI interfaces); as with AppleTalk the goal was simplicity.
Another reason LocalTalk was brilliant, beyond zeroconfig: you could use ADB cables with it! In a pinch, if you needed to get a file from one computer to another, you'd simply detach the cable from your keyboard and use that to connect the two computers together.
> I attempted to use Netscape Navigator and iCab based on the list from here to no avail, Netscape Navigator crashed and iCab reported that it didn't have enough memory
Well the SE is like 1986 technology that predates the web. NCSA Mosaic came out in 1993.
I can't find the min spec for Mac Mosaic 1.x other than "System 7" but you could give it a try. Even Mosaic would have been mostly targeting Macs in wide deployment at the time. Those would be more like Mac II, Centris and Quadra machines with 68030s or 68040s with more RAM and built-in HDDs.
Thanks for the advice. I found a copy of NCSA Mosaic 1.0.3 and it works on the Macintosh SE running System 7! The next version, Mosaic 2.0.1, needs 5MB of free memory. Mosaic does a better job handling larger pages like this blog post, and even chews through the inline SVG diagrams and renders just the text.
My only complaint is that Mosaic uses the serif default font we're all used to, instead of the sans-serif default used by MacWeb, which makes it a little harder to read at such a low resolution.
Added a section on it to the bottom of the article :)
I had a Mac LC with 4MB of RAM when I was a kid. It ran System 7 and was slightly too underpowered to run a web browser. I think you needed at least 6 MB to get Netscape 1.0 to run. Even if you did get it to run it would be painfully slow, especially if you're connected via a 14.4 modem as was common at the time. Loading a single image could take a full minute or more.
Rule of thumb for Z-Modem transfers on a 14.4 modem was 100kB of data per minute.
I'm trying to recall how I did this - maybe RAM disk? Wasn't much of a concern at first, as you could run something like lynx in your dial up terminal and there weren't many images on those early web sites.
“Time to download your storage” is an interesting metric, and it certainly takes longer to fill a terabyte drive from a 1gb/s connection that filling floppies with a modem.
Nice write up! I also have a Mac SE, but I found an Asente Ethernet card and installed that. Then ran ethernet to a small custom pcb that bridges to wireless, all tucked inside the original case.
I have a Mac Plus I've expanded to 4MB and expanded with a 512MB SCSI HD and an Asante SCSI-to-Ethernet adapter by which it's wired to my LAN. I use an FTP programs to browse my Mac Studio via a little FTP server I've got running on it, if I want to install something new on the Mac Plus. Works pretty well.
There's the PiSCSI which is a board that hooks in to a Pi's GPIO pins and acts like a SCSI interface to Macs. I have one and in addition to SCSI drives it can emulate a DynaTalk SCSI/Ethernet interface and use the Pi's wifi to talk to the Internet. I've done this with a Mac SE.
It also will act like a AppleTalk file server using netatalk, it's an easy way to share files with other Macs on your local network.
Would this middleware be relevant with Links [1]? I mean, could we browse heavy-on-js websites such as gmail, facebook ?
[1]: http://links.twibright.com/
I don't have a first-hand account here, but it should work as long as you run Links in graphics mode (but which machine runs Links in graphic mode and not another more enjoyable browser like, say, Netscape?)
This is awesome. I still keep around an ImageWriter II I found in a dumpster two decades ago because they're incredibly reliable and will print no matter what. Ribbons are plentiful and cheap, the printer doesn't care about the kind of paper, and even detailed QR codes and barcodes work fine if you blow them up a bit.
But what's better than a printer that can work with almost any computer made in the last four decades?
I just recently got a LocalTalk card for my ImageWriter II, and between this article and a recent one about custom ROMs for Macs (https://news.ycombinator.com/item?id=37471321), I'm considering having a dedicated m68k machine that can bridge AppleTalk on ethernet with LocalTalk.
I’ve seen some setups for old laser printers (HP LJ IIIp for life!) where a raspberry pie or similar low power computer controls a relay that can turn the printer on and off when not in use, as idle power on those beasts were moderately noticeable.
Which reminds me, I was luck enough to be at one of the last pre-Jobs / transitional era Apple TechTalks in London for dealers. They showed new Stylewriters I recall, which never got released!
Jobs killed all the printers before they shipped, and rightly so!
Although it had a lot of oddities and wasn't the world's most stable program, Ami Pro 3.1 for Windows had hands-down the easiest-to-use equation editor I've ever used. I wasn't doing really complicated math stuff, but was a chem major and the ability to do superscript and subscript directly above one another made any isotope notation very simple.
I seem to recall my school district in the early-to-mid 90's having something like this set up with the very early Macintoshes. The Local Talk cable reminded me of times long forgotten. We could print to any printer in the district from the Chooser.
I made very good use of this wasting paper around the district.
I wonder if a Mac Plus could do this equally well. I am guessing it can. Would be fun to print from the Mac Studio through the Plus to the Imagewriter II.
I run it out of my closet. The blog is a small Go program deployed to a k8s cluster across a few ESXi VMs on an HP DL380 G7, behind a pfSense router also running as a VM and using HAProxy for TLS offloading. My connection is 1gbps symmetric.
With no cache on my local network, Chrome reports the page loads fully in ~500ms, using the "slow 3G connection" performance tab preset it takes ~4s.
If you're nowhere near Little Rock, Arkansas; that might be the issue. I don't yet use a CDN and it always loads the same resolution image ("high" quality JPEG, ~200kb apiece).
It's over-engineered, but I wanted to try running k8s in my homelab to continue learning about networking; deploying a blog seemed like a perfect test case. The cluster uses Traefik, https://github.com/travisghansen/kubernetes-pfsense-controll... to sync service definitions to HAProxy, and MetalLB. The k8s service IPs are routable from my local network, as well as the MetalLB IPs, and MetalLB handles syncing them to pfSense using BGP. pfSense sends DNS for any *.k8s.home.arpa to k8s, so I can resolve e.g. blog.default.svc.k8s.home.arpa on my local network.
One interesting hiccup: I had to introduce a NAT at pfSense because MacBooks don't listen to ICMP redirects, the packet would go laptop -> router -> k8s and then k8s -> laptop. Since the MacBook ignored the ICMP redirect, it drops the reply packet.
In college (Miami University), our computer lab also used PhoneNET (it really simplified wiring) combined with a Cayman Gatorbox to bridge to the campus TCP/IP backbone. By modern standards, painfully slow, but in 1992 the whole university was connected over a single 56K uplink to Columbus, so we hardly noticed.
I have fond memories of playing Spectre with networked opponents over LocalTalk.
To get online from our dorm rooms or offcampus, the only option was dial-up to the university phonebank, a 2400 baud terminal. Using the LocalTalk connection from a lab was much more responsive.