Started 8th February 2023
In the beginning were Teletext, Viewdata, Prestel and Micronet.
By the mid 1980s people had BBC Computers with modems and downloaded software from Micronet. The problem was that Viewdata provided fixed size pages of text, not the binary files of variable length that software typically uses. The solution was a protocol and coding system known as CET, which I believe stood for Council for Educational Technology.
If you want to see a CET downloader, look at the source code for Hearsay (communications software I wrote which features a Viewdata terminal).
Meanwhile, far away, other people where struggling with the problem of sending software over the internet, again the situation is sending a binary file through a channel designed for text only. email attachments in other words. They had a few goes at this e.g. uuencoding which is a base 64 coding - say there are 64 codes you can use safely as text, that's six bits worth. Take four such characters and you have 24 bits, which is 3 bytes worth. Sometimes this is called "4 for 3 coding".
I was supporting such text encoding of binary in my Spark archiver program - often what you would receive would be a uuencoded archive.
To someone exposed to CET as well as the internet coding schemes it was obvious CET was highly inefficient. At some point I had a conversation with Hugo Fiennes (then writing comms software ArcTerm and host software ArcBBS), he told me his new Viewdata software would feature much better downloads. I suggested he would be using binary to text encoding, which he confirmed.
Whilst in general CET is inefficient, for ordinary text it is efficient. As a result taking a binary file, putting it through a base 64 coding program and then using CET on it results in an efficient system. There's no need to throw away CET along with all the terminal software people had.
We passed on this information to Ian Burley, editor of Micronet, and a piece of coding software called FCET+ appeared. The F stands for Fred. Alas I can't recall his second name, at the time I was in contact with him via Micronet.
I added FCET coding and decoding to Spark with the C source code Fred provided. It was a worthwhile improvement because the typical modem was V23, 1200 baud download, 75 baud upload. Maybe 10 seconds to download a Viewdata page.
I come to write this 30 years later because I am removing FCET coding from SparkFS. Back in the mid 80s I would use Viewdata like I use the internet today - reading messages, looking at technical forums, downloading software. However it has all vanished, it as if it never existed. Viewdata pages are compact, 1K a piece. The problem is that discs did not hold much back then and it was not worth keeping stuff. Seemingly no one kept anything.
I do not recall Micronet using archives, but Acorn had a Viewdata system called SID and that did.
You can download the FCET+ code from the link below:
https://www.davidpilling.com/software/fcet.zip