rose-ash/spec/tests/hyperscript-feature-audit.md

_hyperscript Feature Audit

Comprehensive audit of our SX _hyperscript implementation vs the upstream reference.

Implementation files:

• lib/hyperscript/tokenizer.sx — lexer (129 keywords, 17 token types)
• lib/hyperscript/parser.sx — parser (67 internal functions, ~3450 lines)
• lib/hyperscript/compiler.sx — compiler (AST to SX, ~100 dispatch cases)
• lib/hyperscript/runtime.sx — runtime shims (49 functions)

Test files:

• spec/tests/test-hyperscript-behavioral.sx — 381 conformance tests (from upstream)
• spec/tests/test-hyperscript-conformance.sx — 184 additional conformance tests
• spec/tests/test-hyperscript-tokenizer.sx — 43 tokenizer tests
• spec/tests/test-hyperscript-parser.sx — 93 parser tests
• spec/tests/test-hyperscript-compiler.sx — 44 compiler tests
• spec/tests/test-hyperscript-runtime.sx — 26 runtime tests
• spec/tests/test-hyperscript-integration.sx — 12 integration tests

Upstream reference: spec/tests/hyperscript-upstream-tests.json — 831 tests from upstream master+dev

---

Upstream Test Breakdown

Complexity	Count	Description
simple	469	DOM-based tests, simplest to translate
run-eval	83	Eval-only tests (no DOM setup)
evaluate	125	Full browser eval with DOM interaction
promise	57	Async/promise-based tests
eval-only	39	Pure expression evaluation
script-tag	36	Tests using <script type="text/hyperscript">
sinon	17	Tests requiring sinon mock (fetch)
dialog	5	Dialog-specific tests

Of the 469 simple tests, 454 are "clean" (no [@, ${, {css}, or <sel/> patterns that our tokenizer doesn't handle).

Our 381 behavioral tests were generated from the simple upstream tests and represent features that parse + compile + execute correctly in our sandbox environment.

---

Feature-by-Feature Audit

Legend

• IMPL+TEST = Implemented in all four layers (tokenizer/parser/compiler/runtime) AND tested
• PARTIAL = Compiles but not all sub-features work, or only basic forms tested
• NOT IMPL = Parser/compiler doesn't handle it at all
• IMPL-UNTESTED = Code exists in implementation but no test coverage

---

COMMANDS

Core assignment/mutation

Feature	Status	Tests	Notes
set ... to ...	IMPL+TEST	24+	Properties, locals, globals, attrs, styles, indirect. parse-set-cmd + emit-set
put ... into/before/after ...	IMPL+TEST	37+	Full positional insertion. parse-put-cmd + emit-set/hs-put!
get	IMPL+TEST	5	Parsed as expression (property access); call-cmd dispatch also handles get
increment	IMPL+TEST	20	Variables, attributes, properties, arrays, possessives, style refs. parse-inc-cmd + emit-inc
decrement	IMPL+TEST	20	Mirror of increment. parse-dec-cmd + emit-dec
append ... to ...	IMPL+TEST	13	parse-append-cmd -> dom-append
default	IMPL+TEST	9	Array elements, style refs, preserves zero/false. (Tested in behavioral)
empty/clear	IMPL+TEST	12	Elements, inputs, textareas, checkboxes, forms. (Tested in behavioral)
swap	IMPL+TEST	4	Variable/property/array swaps. (Tested in behavioral)

Class manipulation

Feature	Status	Tests	Notes
add .class	IMPL+TEST	14+	Single, multiple, double-dash, colons. parse-add-cmd -> dom-add-class
add .class to <target>	IMPL+TEST	14+	Target resolution for classes
add [@attr="val"]	NOT IMPL	0	Tokenizer doesn't emit [@ as attribute-set token. 4 upstream tests skipped
add {css-props}	NOT IMPL	0	CSS property block syntax not tokenized. 2 upstream tests skipped
remove .class	IMPL+TEST	10+	parse-remove-cmd -> dom-remove-class
remove (elements)	IMPL+TEST	5	Remove self, other, parent elements
remove [@attr]	NOT IMPL	0	Same tokenizer limitation as add [@]
remove {css}	NOT IMPL	0	CSS block removal not implemented
toggle .class	IMPL+TEST	28+	Single, multiple, timed, between two classes. parse-toggle-cmd -> hs-toggle-class!/hs-toggle-between!
toggle .class for <duration>	IMPL+TEST	1	Timed toggle
toggle .class until <event>	IMPL+TEST	1	Event-gated toggle
toggle between .a and .b	IMPL+TEST	1	hs-toggle-between! runtime function
toggle [@attr]	NOT IMPL	0	Attribute toggle not implemented
toggle {css}	NOT IMPL	0	CSS block toggle not implemented
take .class	IMPL+TEST	12	From siblings, for others, multiple classes. parse-take-cmd -> hs-take!
take [@attr]	IMPL+TEST	10	Attribute take from siblings. (Tested in behavioral)

Control flow

Feature	Status	Tests	Notes
if ... then ... end	IMPL+TEST	18+	With else, else if, otherwise, triple nesting. parse-if-cmd
if ... else ...	IMPL+TEST	18+	Naked else, else end, multiple commands
repeat ... times ... end	IMPL+TEST	23+	Fixed count, expression count, forever, while, for-in. parse-repeat-cmd + hs-repeat-times/hs-repeat-forever
repeat forever	IMPL+TEST	1+	hs-repeat-forever
repeat while	IMPL+TEST	1+	While condition in repeat mode
repeat for x in collection	IMPL+TEST	5+	For-in loop mode
for x in collection ... end	IMPL+TEST	5+	parse-for-cmd + emit-for -> for-each
for x in ... index i	IMPL+TEST	2	Index variable support
return	IMPL+TEST	varies	parse-return-cmd, bare and with expression
throw	IMPL+TEST	varies	parse-throw-cmd -> raise
catch	IMPL+TEST	14	Exception handling in on blocks. (Tested in behavioral)
finally	IMPL+TEST	6	Finally blocks. (Tested in behavioral)
break	PARTIAL	0	Keyword recognized by tokenizer, but no dedicated parser/compiler path
continue	PARTIAL	0	Keyword recognized by tokenizer, but no dedicated parser/compiler path
unless	PARTIAL	0	Keyword recognized but no dedicated parser path (falls through)

Async/timing

Feature	Status	Tests	Notes
wait <duration>	IMPL+TEST	7	Duration parsing (ms, s). parse-wait-cmd -> hs-wait using perform
wait for <event>	IMPL+TEST	7	Wait for DOM event. hs-wait-for using perform
wait for <event> from <source>	IMPL+TEST	1	Source-specific event wait
wait for <event> or <timeout>	IMPL+TEST	2	Timeout variant
settle	IMPL+TEST	1	hs-settle using perform. Compiler emits (hs-settle me)
Async transparency	IMPL+TEST	varies	perform/IO suspension provides true pause semantics

Events/messaging

Feature	Status	Tests	Notes
send <event>	IMPL+TEST	8	parse-send-cmd -> dom-dispatch. Dots, colons, args
send <event> to <target>	IMPL+TEST	8	With detail dict, target expression
trigger <event>	IMPL+TEST	varies	parse-trigger-cmd -> dom-dispatch

Navigation/display

Feature	Status	Tests	Notes
go to <url>	IMPL+TEST	3	parse-go-cmd -> hs-navigate!
hide	IMPL+TEST	14	Multiple strategies (display:none, opacity:0, visibility:hidden). Custom strategies. (Tested in behavioral)
show	IMPL+TEST	2	parse-show-cmd. (Tested in behavioral)
transition ... to ... over ...	IMPL+TEST	22	Properties, custom duration, other elements, style refs. parse-transition-cmd + hs-transition
log	IMPL+TEST	4	parse-log-cmd -> console-log
halt	IMPL+TEST	6	Event propagation/default prevention. (Tested in behavioral)
halt the event	IMPL+TEST	2	Stops propagation, continues execution
halt bubbling	IMPL+TEST	1	Only stops propagation
halt default	IMPL+TEST	1	Only prevents default

Function/behavior

Feature	Status	Tests	Notes
call fn(args)	IMPL+TEST	5	parse-call-cmd, global and instance functions
call obj.method(args)	IMPL+TEST	varies	Method call dispatch via hs-method-call
def fn(params) ... end	IMPL+TEST	3	parse-def-feat -> define
behavior Name(params) ... end	IMPL+TEST	varies	parse-behavior-feat + emit-behavior
install BehaviorName	IMPL+TEST	2	parse-install-cmd -> hs-install
make a <Type>	IMPL+TEST	varies	parse-make-cmd + emit-make -> hs-make. Called keyword support
render <component>	IMPL+TEST	varies	parse-render-cmd with kwargs, position, target. Bridges to SX component system

DOM/IO

Feature	Status	Tests	Notes
fetch <url>	IMPL+TEST	6	parse-fetch-cmd -> hs-fetch. JSON/text/HTML formats. URL keyword deprecated but parsed
fetch ... as json/text/html	IMPL+TEST	6	Format dispatch in runtime
measure	IMPL+TEST	varies	parse-measure-cmd -> hs-measure using perform
focus	NOT IMPL	0	No parser, 3 upstream tests (all evaluate complexity)
scroll	NOT IMPL	0	No parser, 8 upstream tests (all evaluate complexity)
select	NOT IMPL	0	No parser, 4 upstream tests (all evaluate complexity)
reset	IMPL+TEST	8	Forms, inputs, checkboxes, textareas, selects. (Tested in behavioral)
morph	IMPL+TEST	4	(Tested in behavioral, simple complexity)
dialog (show/open/close)	IMPL+TEST	5	Modal dialogs, details elements. (Tested in behavioral)

Other commands

Feature	Status	Tests	Notes
tell <target> ... end	IMPL+TEST	10	parse-tell-cmd. Scoping (you/your/yourself), attribute access, me restoration. (Tested in behavioral)
js ... end	PARTIAL	1	Keyword recognized, parse-atom handles eval keyword -> sx-eval, but inline JS blocks not fully supported
pick	NOT IMPL	0	7 upstream tests (all eval-only complexity). No parser path
beep!	IMPL+TEST	1	Debug passthrough. parse-atom recognizes beep!, runtime hs-beep is identity

---

FEATURES (Event handlers / lifecycle)

Feature	Status	Tests	Notes
on <event> ... end	IMPL+TEST	59+	parse-on-feat + emit-on -> hs-on. Dots, colons, dashes in names
on <event> from <source>	IMPL+TEST	5+	Source-specific listeners
every <event>	IMPL+TEST	1+	hs-on-every — no queuing
on ... [<filter>]	IMPL+TEST	3+	Event filtering in on blocks
Event destructuring	IMPL+TEST	1+	can pick detail/event properties
on <event> count N / range	IMPL+TEST	4	Count filter, range filter, unbounded range
on mutation	IMPL+TEST	10	Attribute, childList, characterData mutations. Cross-element. (Tested in behavioral)
on first <event>	IMPL+TEST	1	One-shot handler
on load	IMPL+TEST	1	Load pseudo-event
Queue modes (queue, first, last, all, none)	IMPL+TEST	5	Event queuing strategies
init ... end	IMPL+TEST	1+	parse-init-feat -> hs-init
def name(params) ... end	IMPL+TEST	3+	Feature-level function definitions
behavior Name(params) ... end	IMPL+TEST	varies	Feature-level behavior definition
on <event> debounce <dur>	NOT IMPL	0	Debounce modifier not parsed
on <event> throttle <dur>	NOT IMPL	0	Throttle modifier not parsed
connect	NOT IMPL	0	No parser path
disconnect	NOT IMPL	0	No parser path
worker	NOT IMPL	0	No parser path
socket	NOT IMPL	0	4 upstream tests (all eval-only). No parser path
bind	PARTIAL	1	Keyword in tokenizer. 44 upstream tests (mostly promise/evaluate). 1 simple test in behavioral. Parser doesn't have dedicated bind command
when (reactive)	IMPL+TEST	5	Reactive when handler. (Tested in behavioral)
live	NOT IMPL	0	23 upstream tests (evaluate/promise). No parser path
resize	NOT IMPL	0	3 upstream tests (evaluate). No parser path
intersect	NOT IMPL	0	No upstream tests. No parser path
every N seconds (polling)	NOT IMPL	0	Time-based polling pseudo-feature not parsed

---

EXPRESSIONS

Literals & references

Feature	Status	Tests	Notes
Number literals	IMPL+TEST	Yes	Integer and float
String literals	IMPL+TEST	Yes	Single and double quoted
Boolean literals (true/false)	IMPL+TEST	Yes
null/undefined	IMPL+TEST	Yes	Both produce (null-literal)
me/I/my	IMPL+TEST	Yes	Self-reference. my triggers possessive tail
it/its	IMPL+TEST	Yes	Result reference. its triggers possessive tail
event	IMPL+TEST	Yes	Event object reference
target	IMPL+TEST	Yes	event.target
detail	IMPL+TEST	Yes	event.detail
sender	IMPL+TEST	Yes	Event sender reference
result	IMPL+TEST	Yes	Implicit result
the	IMPL+TEST	Yes	Article prefix, triggers parse-the-expr
you/your/yourself	IMPL+TEST	Yes	Tell-scoping references
Local variables (:name)	IMPL+TEST	Yes	Tokenizer emits local type
Template literals	IMPL+TEST	Yes	${expr} and $ident interpolation. Compiler handles nested parsing
Array literals [a, b, c]	IMPL+TEST	Yes	parse-array-lit
Object literals {key: val}	IMPL+TEST	Yes	parse-atom -> object-literal
Block literals \ param -> expr	IMPL+TEST	Yes	Lambda syntax in parse-atom

Property access

Feature	Status	Tests	Notes
Dot notation (obj.prop)	IMPL+TEST	Yes	parse-prop-chain. Chained access
Method calls (obj.method(args))	IMPL+TEST	Yes	parse-prop-chain + method-call AST node
Bracket access (arr[i])	IMPL+TEST	Yes	parse-poss handles bracket-open -> array-index
Array slicing (arr[i..j])	IMPL+TEST	Yes	array-slice AST node -> hs-slice
Possessive (obj's prop)	IMPL+TEST	Yes	parse-poss + parse-poss-tail
of syntax (prop of obj)	IMPL+TEST	Yes	In parse-cmp -> (of ...) AST
Attribute ref (@attr)	IMPL+TEST	Yes	Tokenizer emits attr type. Compiler -> dom-get-attr/dom-set-attr
Style ref (*prop)	IMPL+TEST	Yes	Tokenizer emits style type. Compiler -> dom-get-style/dom-set-style
Class ref (.class)	IMPL+TEST	Yes	Tokenizer emits class type
ID ref (#id)	IMPL+TEST	Yes	Tokenizer emits id type -> (query "#id")
Selector ref (<sel/>)	IMPL+TEST	Yes	Tokenizer emits selector type -> (query sel). 8 upstream simple tests use this
[@attr="val"] set syntax	NOT IMPL	0	Tokenizer doesn't handle [@ — attribute SET inside add/remove/toggle

Comparison operators

Feature	Status	Tests	Notes
is / is not	IMPL+TEST	Yes	parse-cmp. Equality and negation
is equal to / is not equal to	IMPL+TEST	Yes	Strict equality
is really / is not really	IMPL+TEST	Yes	type-check-strict / strict type check
is a/an <Type>	IMPL+TEST	Yes	Type checking with a/an article
is not a/an <Type>	IMPL+TEST	Yes	Negated type check
is empty / is not empty	IMPL+TEST	Yes	hs-empty? runtime
exists / does not exist	IMPL+TEST	Yes	exists? AST node
matches / does not match	IMPL+TEST	Yes	hs-matches? runtime
contains / does not contain	IMPL+TEST	Yes	hs-contains? runtime
includes / does not include	IMPL+TEST	Yes	Aliases for contains
<, >, <=, >=	IMPL+TEST	Yes	Standard operators in parse-cmp and parse-arith
less than / greater than	IMPL+TEST	Yes	English word forms
less than or equal to	IMPL+TEST	Yes	Full English form
greater than or equal to	IMPL+TEST	Yes	Full English form
==, !=	IMPL+TEST	Yes	Op tokens in parse-cmp
===, !==	IMPL+TEST	Yes	Strict equality ops -> strict-eq
between	PARTIAL	0	Keyword recognized in tokenizer but no dedicated parser path in parse-cmp
starts with / ends with	NOT IMPL	0	No parser path
precedes / follows	NOT IMPL	0	No parser path
is <prop> (property truthiness)	IMPL+TEST	Yes	prop-is AST -> hs-prop-is

Logical operators

Feature	Status	Tests	Notes
and	IMPL+TEST	Yes	parse-logical
or	IMPL+TEST	Yes	parse-logical
not	IMPL+TEST	Yes	parse-atom prefix
no	IMPL+TEST	Yes	parse-atom prefix -> (no expr) -> hs-falsy?

Math operators

Feature	Status	Tests	Notes
+, -, *, /	IMPL+TEST	Yes	parse-arith
% (modulo)	IMPL+TEST	Yes	parse-arith handles % and mod keyword -> modulo
mod	IMPL+TEST	Yes	Keyword alias for %
Unary -	IMPL+TEST	Yes	In parse-atom
CSS unit postfix (10px, 50%)	IMPL+TEST	Yes	string-postfix AST node

String operations

Feature	Status	Tests	Notes
split by	IMPL+TEST	Yes	parse-collection -> coll-split -> hs-split-by
joined by	IMPL+TEST	Yes	parse-collection -> coll-joined -> hs-joined-by

Type coercion

Feature	Status	Tests	Notes
as String	IMPL+TEST	Yes	parse-cmp handles as keyword -> hs-coerce
as Int / as Float	IMPL+TEST	Yes	Numeric coercion
as Array	IMPL+TEST	Yes	Collection coercion
as Object	IMPL+TEST	Yes	Object coercion
as JSON	IMPL+TEST	Yes	JSON serialization/parse
as HTML / as Fragment	IMPL+TEST	Yes	HTML/DOM coercion
as Date	IMPL+TEST	Yes	Date coercion
as Number	IMPL+TEST	Yes	Number coercion
as Values	IMPL+TEST	Yes	Form values coercion
Custom type coercion (:param)	IMPL+TEST	Yes	as Type:param syntax parsed
as response	IMPL+TEST	1	(Tested in behavioral fetch tests)

Positional / traversal expressions

Feature	Status	Tests	Notes
first	IMPL+TEST	Yes	parse-pos-kw -> hs-query-first / hs-first
last	IMPL+TEST	Yes	parse-pos-kw -> hs-query-last / hs-last
first ... in ...	IMPL+TEST	Yes	Scoped first
last ... in ...	IMPL+TEST	Yes	Scoped last
next	IMPL+TEST	Yes	parse-trav -> hs-next. Class, ID, wildcard selectors
previous	IMPL+TEST	Yes	parse-trav -> hs-previous
closest	IMPL+TEST	Yes	parse-trav -> dom-closest. (Tested in behavioral)
random	PARTIAL	0	Keyword recognized but no dedicated parser/compiler path

Collection expressions

Feature	Status	Tests	Notes
where <condition>	IMPL+TEST	Yes	parse-collection -> coll-where -> filter with it binding
sorted by <key>	IMPL+TEST	Yes	parse-collection -> coll-sorted -> hs-sorted-by
sorted by <key> descending	IMPL+TEST	Yes	coll-sorted-desc -> hs-sorted-by-desc
mapped to <expr>	IMPL+TEST	Yes	parse-collection -> coll-mapped -> map
split by <sep>	IMPL+TEST	Yes	In parse-collection
joined by <sep>	IMPL+TEST	Yes	In parse-collection
some x in coll with <pred>	IMPL+TEST	Yes	Quantifier in parse-atom -> (some ...)
every x in coll with <pred>	IMPL+TEST	Yes	Quantifier in parse-atom -> (every ...)
filter (standalone)	NOT IMPL	0	No standalone filter command
reduce	NOT IMPL	0	No reduce in collection expressions
in (membership)	IMPL+TEST	Yes	is in / is not in in parse-cmp -> in? / not-in?

Special forms

Feature	Status	Tests	Notes
eval / SX interop	IMPL+TEST	Yes	parse-atom handles eval -> (sx-eval ...). Inline SX from parens or expression
Component refs (~name)	IMPL+TEST	Yes	Tokenizer emits component type. Compiler resolves to SX component call
new keyword	PARTIAL	0	Keyword recognized but no dedicated constructor path

---

FEATURES NOT IMPLEMENTED (by upstream category)

These upstream test categories have zero coverage in our implementation:

Category	Upstream Tests	Complexity	Why Missing
askAnswer	5	dialog	ask/answer dialog commands not parsed
asExpression	17	eval-only/run-eval	as expression standalone evaluation — partially covered by as in comparisons
asyncError	2	evaluate/promise	Async error propagation edge cases
attributeRef	1	evaluate	@attr as standalone assignable
cookies	1	eval-only	Cookie access not implemented
evalStatically	8	eval-only	Static evaluation optimization
focus	3	evaluate	focus command not implemented
in	1	run-eval	Standalone in expression
live	23	evaluate/promise	live event sources not implemented
logicalOperator	3	eval-only	Standalone logical operator eval (covered by inline use)
mathOperator	5	run-eval	Standalone math eval (covered by inline use)
measure	2	evaluate	measure runtime needs real DOM
objectLiteral	1	run-eval	Standalone object literal eval (implemented, just no dedicated run-eval test)
pick	7	eval-only	pick command not parsed
queryRef	1	evaluate	<sel/> standalone (implemented but test requires DOM)
reactive-properties	4	evaluate/promise/run-eval	Reactive property observation
relativePositionalExpression	4	eval-only/evaluate	Relative position expressions (next/previous as standalone)
resize	3	evaluate	resize pseudo-event not implemented
scroll	8	evaluate	scroll command not implemented
select	4	evaluate	select command not implemented
settle	1	simple	settle command exists but upstream test is DOM-dependent
socket	4	eval-only	WebSocket feature not implemented
splitJoin	7	run-eval	Split/join standalone eval (implemented, tests are run-eval complexity)

---

DOM-SCOPE (^var) — Extended feature

Our implementation includes the full dom-scope (^var) feature:

Feature	Status	Tests
^var read from ancestor	IMPL+TEST	23
^var write propagates to ancestor	IMPL+TEST	23
isolated stops resolution	IMPL+TEST	1
closest ancestor wins (shadowing)	IMPL+TEST	1
when reacts to ^var changes	IMPL+TEST	2
bind with ^var	IMPL+TEST	1

---

COMPONENT feature (web components)

Feature	Status	Tests
<template> component registration	IMPL+TEST	14
#if conditionals in templates	IMPL+TEST	2
Named and default slots	IMPL+TEST	2
Component ^var isolation	IMPL+TEST	1
Multiple independent instances	IMPL+TEST	1

---

Summary Statistics

Category	Count
Features fully implemented + tested	~55
Features partially implemented	~8
Features not implemented	~18
Total upstream tests	831
Tests translated to SX (behavioral)	381 (46%)
Additional SX conformance tests	184
Tests skippable (non-simple complexity)	362 (44%)
Simple tests blocked by HTML patterns	15 (2%)
Clean simple tests available	454
Gap: clean simple not yet translated	~73

What blocks the remaining 73 clean simple tests

These tests exist in clean simple upstream but are not in our 381 behavioral tests. They likely involve features that:

1. Require real DOM interaction (hide with strategies, fetch with network)
2. Were added to upstream after our test generation
3. Involve categories we partially support (halt, dialog, reset, morph, liveTemplate)

Top priorities for implementation

1. [@attr="val"] syntax (tokenizer) — 4 simple upstream tests blocked
2. {css-props} block syntax (tokenizer) — 2 simple upstream tests blocked
3. debounce/throttle event modifiers — Common real-world usage
4. scroll command — 8 upstream tests
5. focus command — 3 upstream tests
6. select command — 4 upstream tests
7. pick command — 7 upstream tests
8. live feature — 23 upstream tests, key for reactive data
9. between comparison — Keyword exists, needs parser/compiler path
10. starts with/ends with — Common string comparisons