| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Two columns: edit code on left and run code on right.
In both columns, the action is at the top of the screen, and stuff
gradually flows downward unless pulled back up.
Stuff you run on the right side will eventually become reproducible test
cases that runs constantly on every change.
We also have a tentative signature for the 'edit' routine: it'll take an
input string and return an output string when it finishes.
More columns expand to the right. Number of columns shown will remain
fixed at any time depending on screen width. But for now we support
exactly two columns.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Yet another repl bugfix.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
I'd hoped finding that off-by-one error would fix the other intermittent
issues I've been seeing (commit 1512), but no they're still there.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Variable 'instruction_counter' was obfuscating more than it clarified.
|
| |
|
|
|
|
|
|
| |
Extracting a function for the prompt before starting to test it.
Still not clear how I should proceed. Need to express screen dependency
to 'run-interactive' somehow. Supporting generics might finally be on
the critical path.
|
| | |
|
| |
|
|
|
| |
But we really should test the top-level integration with
'run-interactive'.
|
| |
|
|
| |
Was missing a newline. More brittleness.
|
| |
|
|
|
|
| |
Just figured out why a first keystroke of backspace was sending me out
for a spin: run_interactive needs all early exits that don't actually
run anything to increment the current_step_index(). FML, this is lousy..
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
In the process we now support unicode in all mu strings!
|
| |
|
|
|
|
|
|
| |
I'm still having trouble with prematurely closed strings or strings
never being closed. Remember that when escaping brackets inside
scenarios you usually need two levels of escaping. The single-level is
a no-op and usually misleading. When brackets balance, don't bother
escaping. When they don't, escape twice.
|
| |
|
|
|
|
|
|
|
|
| |
As expected, I've already found two bugs in the repl during manual
testing. Still don't have a good characterization of when things go off
the rails.
But certainly hitting backspace too many times isn't messing things up
as I'd feared. Which might also be a bug, if the program isn't behaving
as I expect.
|
| |
|
|
|
|
|
| |
But I'm sure the next bug is nearby. And even if there's a solution
around here, the logic is now quite convoluted, with strings sometimes
returning on completion and sometimes recursing further in. We'll need
to rethink this.
|
| |
|
|
|
| |
slurp-string should sometimes return and sometimes recurse. Graah.
But the trace browser is awesome.
|
| |
|
|
|
| |
Ah, the failing test isn't what I thought. This is a regression.
Let's debug this using the trace-browser! Little tweak as I use it.
|
| |
|
|
| |
Aaalmost there..
|
| |
|
|
|
|
| |
More groundwork for dealing with backspacing back into string: instead
of simply exiting the current stack frame when the string completes,
create a *new* stack frame. More tests failing, but we're getting closer.
|
| | |
|
| |
|
|
| |
Fix messages on failed screen tests.
|
| |
|
|
|
| |
repl still needs to handle backspacing *back into* strings. This tries
to lay the groundwork.
|
| | |
|
| | |
|
| | |
|