The Official Aztec C Online Museum (2009)

I've always felt the the 6502 architecture wasn't really amenable to C (mainly due to the limited stack). A quick glance at the site with an eye toward accommodations made for the platform turned up the interesting hybrid architecture of the compiler suite. You can see it detailed in the section "1.6 C65 and CCI, The Speed Versus Size Dilemma"[0].

Basically, there's an optional compiler that emits a p-code (rather than 6502 machine code) that's interpreted by a VM to make 16 and 32-bit operations more space efficient. This is highly reminiscent of SWEET16[1]. Also makes me think of the Apollo Guidance Computer and the "Interpreter"[2].

This looks like a lot of fun. I'd like to get this running on my Apple IIgs!

[0] http://aztecmuseum.ca/docs/AztecC_minimanual.txt [1] https://en.wikipedia.org/wiki/SWEET16 [2] https://en.wikipedia.org/wiki/Apollo_Guidance_Computer#Softw...

I have very fond memories of my time there, I joined not too long out of high school in the late 80's. I transitioned from QA to development while learning C, learning how to program a few different assemblies, putting together a TSR (!) library. It was a great little company that changed my life. I still use VI keybindings because everyone internally used the Aztec version of VI :-)

It was very cool that they built a mini Unix toolset as part of the compiler suite so you had kinda-sorta unix on DOS, Mac OS MPW shell, Amiga....

It was a rare privilege for a 20 year old in the 80s to see code that beautiful and tight.

I had a good use for these old compilers- in this case I ended up using code from the Zortech-C library to reduce memory usage for ARM embedded. It turns out that the float to ascii conversion functions for printf from the Zortech-C library are much smaller than the ones in newlib. I'm sure they are less accurate, but I don't care for this application. Same for strtod. Also they are faster.

I put the code here:

https://github.com/jhallen/joes-sandbox/blob/master/snippets...

I see the Aztec ones for x86 are hand assembly..

Hmm... I thinks this is what I used to learn C - on an Amiga.

Oh, the memories...

... most of them about waiting for the disk to finish reading just to run out of memory thanks to Intuitions.h being such a beast. I think one of my first real C programs was a pre-processor and minimiser for the amiga UI header files.

I recently started working on a single-board Z80-based computer, running CP/M. I wrote a couple of simple programs in Z80 assembly, including a text-based adventure game, and other utilities.

It was so cool to be able to run the Aztec compiler in-situ, on the computer itself, and I'm glad the official sources have been preserved.

There are a lot of online archives of random .zip files for CP/M, but few are as well documented as the Aztec collection.

The thing I liked about Aztec C was that they gave you the source code for their libc. You could easily write and integrate your own I/O drivers for stdin and stdout. It really made embedded C development easy.

I worked at this company as a support engineer in the summer between high school and college. It was a blast.

Does anyone have numbers on how the backend compares to CC65 for performance on the 6502?

This was the first compiler I paid for as a grad student circa 1986. It was wonderfully useful, well documented. Tech support always responded to my letters and emails. And they sent me updates for years. I loved Aztec C.

Ahh, the memories. Aztec C on an Apple II with a giant Corvus 5 MB drive (whoosh, whoosh, whoosh goes the drive head). It’s where I learned “:wq”, thanks to Z!