diff options
| author | user <user@clank> | 2026-08-15 00:02:51 +0200 |
|---|---|---|
| committer | user <user@clank> | 2026-08-15 00:02:51 +0200 |
| commit | 599da705d93f27efb2a4c62b0747356d97e8f9b3 (patch) | |
| tree | 34dca7208797dba5088c52abdf328095b33fde90 /tests | |
| parent | contrib: draw the shadertoy preview at the size it means (diff) | |
| download | slosh-599da705d93f27efb2a4c62b0747356d97e8f9b3.tar.gz slosh-599da705d93f27efb2a4c62b0747356d97e8f9b3.tar.xz slosh-599da705d93f27efb2a4c62b0747356d97e8f9b3.zip | |
contrib: presets worth stealing, and a test that they do anything
Eleven examples, grouped, and most of them are less "effect" than
*instrument*: a cursor line, a crosshair, indent guides, a right margin at
three quarters, a torch that dims everything more than eight cells from where
you are looking. Things that tell you where you are, inside a program that
has never heard of any of them -- which is the argument for shading contents
rather than chrome, made in examples instead of prose.
Then the fun: phosphor and amber CRTs (grayscale, tint, scanlines, vignette
as a four-pass chain), film grain, a rolling bar, an interlace flicker; sonar
pings expanding from the cursor, ripples, a barber pole, plasma from two
triangle waves, matrix rain with a fading tail; and every built-in written as
an expression, because `vignette` being `dim` with an argument is the whole
design and is better shown than argued.
Two rules the presets follow, both about staying identical to the C: never
multiply `t` (it is a millisecond clock and gets large -- divide it down
first and the arithmetic stays in an int32 forever), and keep hash
coefficients small for the same reason. `(x * 11) % 23` is fine; the usual
Knuth constant is a different number in C than in JavaScript.
The test now lifts the *presets* out of the page as well as the evaluator, so
one added later is conformance-checked without anyone remembering to add it.
And it evaluates every preset over a whole pane at four points in time and
fails any that is zero everywhere -- an example that does nothing reads as a
broken shader system rather than a bad example.
That check caught its own first draft: sampling every other row made three
`y % 2` effects look dead when they were fine. It samples every cell now,
which costs six hundredths of a second for four hundred thousand
evaluations, because the thing being tested is quick.
Diffstat (limited to 'tests')
| -rw-r--r-- | tests/test_shadertoy.py | 51 |
1 files changed, 38 insertions, 13 deletions
diff --git a/tests/test_shadertoy.py b/tests/test_shadertoy.py index ce508b1..0b1dd49 100644 --- a/tests/test_shadertoy.py +++ b/tests/test_shadertoy.py @@ -25,6 +25,7 @@ from harness import check, report HERE = os.path.dirname(os.path.abspath(__file__)) HTML = os.path.join(HERE, "..", "contrib", "shadertoy.html") EVAL = os.path.join(HERE, "..", "build", "expr_eval") +SCRIPT = open(HTML).read().split("<script>")[1].split("</script>")[0] # Environments to evaluate at: corners, the middle, and a moved cursor. ENVS = [ @@ -34,19 +35,23 @@ ENVS = [ dict(x=40, y=12, cols=100, rows=40, curx=99, cury=39, cursor=1, focused=1, t=7), ] +def page_presets(): + """Every preset the page ships, lifted out of the page. + + Read rather than listed, so a preset added later is checked without anyone + remembering to add it here -- which nobody would.""" + prefix = SCRIPT[:SCRIPT.index("/* ---- the chain UI")] + out = subprocess.run(["node", "-e", prefix + "\nconsole.log(JSON.stringify(PRESETS));"], + capture_output=True, text=True) + groups = json.loads(out.stdout) + return [(f"{group}/{name}", i, p["amount"]) + for group, entries in groups.items() + for name, chain in entries.items() + for i, p in enumerate(chain)] + + EXPRS = [ - # the presets the page ships, which are the ones people will actually copy - "(y % 2) * 40", - "(x > cols - 10) * 120", - "clamp(dist(x, y, curx, cury) * 9, 0, 200)", - "clamp(dist(x, y, cols / 2, rows / 2) * 4, 0, 190)", - "(x == 40) * 220", - "((y / 3) % 2) * 60", - "(((x / 8) + (y / 4)) % 2) * 45", - "clamp((t / 6) % 300 - 45, 0, 160)", - "clamp(abs(x - (t / 24) % cols) * 22, 0, 200)", - "180", - # and the corners of the language + # the corners of the language "1 + 2 * 3", "(1 + 2) * 3", "10 - 3 - 2", @@ -146,7 +151,9 @@ def main(): check("the expression evaluator is built", False, EVAL) return report() - exprs = EXPRS + rand_exprs(40) + presets = page_presets() + check("the page ships presets to check", len(presets) >= 20, str(len(presets))) + exprs = EXPRS + [e for _, _, e in presets] + rand_exprs(40) pairs = [(env, e) for e in exprs for env in ENVS] import tempfile @@ -177,6 +184,24 @@ def main(): not bad, "\n".join(f" {e!r} at {env} -> C {w}, JS {g}" for e, env, w, g in bad[:8])) + # A preset that computes 0 everywhere is a preset that does nothing, which + # is a broken example rather than a subtle one: it looks like the shader + # system is not working. Checked over a whole pane and four points in time, + # cell by cell -- sampling every other row made three `y % 2` effects look + # dead when they were fine. + grid = [dict(x=x, y=y, cols=64, rows=24, curx=30, cury=12, cursor=1, + focused=1, t=t) + for t in (0, 500, 1500, 4000) for y in range(24) for x in range(64)] + flat = [(env, e) for _, _, e in presets for env in grid] + vals = c_values(flat) + per = len(grid) + for idx, (label, i, expr) in enumerate(presets): + window = vals[idx * per:(idx + 1) * per] + nums = [int(v) for v in window if not v.startswith("error")] + check(f"{label} pass {i} does something", + nums and max(0, min(255, max(nums))) > 0, + f"{expr!r} is 0 everywhere") + # And that a refusal is a refusal on both sides, since a page that quietly # accepts what the config rejects teaches the wrong language. refused = ["nosuchvar + 1", "wobble(1, 2)", "(1 + 2", "1 +", "1 + 2 )", "1 ? 2"] |
