I've wrestled for a long time with how to support integer division with
its hard-coded registers. The answer's always been staring me in the face:
just turn it into a function! We already expect function outputs to go
to hard-coded registers.
Minor tweaks to get Mu shell running nicely on a Linux console atop Qemu.
We also need to switch a few 256-color codes to 8-color mode. I'm not
sure whether/how to patch the repo for those.
Both manual tests described in commit 7222 now work.
To make them work I had to figure out how to copy a file. It
requires a dependency on a new syscall: lseek.
Ok, I found a failing manual test for files as well.
Here are the two steelman tests, one for screens and one for files:
1.
5 5 fake-screen =s
s 1 down 1 right
ctrl-d foo
expand
final state:
s foo foo
s 1 down 1 right ⇗
┌─────┐ ┌─────┐
┌─────┐ 1 ┌─────┐ 1 ┌─────┐ │
┌─────┐ │ ┌─────┐ │ ─
│ │
│ │
─
└─────┘ └─────┘
└─────┘ └─────┘ └─────┘
└─────┘ └─────┘
2.
"x" open =f
f read f read
ctrl-d read2
expand
final state:
f read2 read2
f read f read ⇗
FILE ❝def❞
FILE ❝abc❞ FILE ❝❞
❝def❞ ❝ghi❞
In both cases there are 3 levels of issues:
- getting a single-line expression to work
- getting a single-line expression to work when operating on a binding
defined in a previous line
- getting an expanded function call to work
The third is where the rub is right now. And what both examples above share
is that the function performs 2 mutations to the screen/file.
So we need a deep copy after all. And it's not very clear how to copy a
file descriptor including the seek location. Linux's dup() syscall creates
an alias to the file descriptor. And opening /proc seems awfully Linux-specific:
https://stackoverflow.com/questions/54727231/duplicating-file-descriptor-and-seeking-through-both-of-them-independently/54727424#54727424
This bug was incredibly painful to track down: the one-line fix is to replace
'line' with 'first-line' in the call to 'evaluate' in render-line before
recursing.
Things that made it challenging:
- A high degree of coiling with recursive calls and multiple places of
evaluation.
- An accidental aliasing in bindings (when rendering the main column in
render-line) that masked the underlying bug and made things seem to work
most of the time.
- Too many fucking arguments to render-line, a maze of twisty line objects
all alike.
Kartik Agaram