FnAddr (Fnval n) emitted the symbol, which is right for a whole-program build
and wrong the instant anything is redefined into it. It now reads the cell,
and so does every direct call, which is what emit.ml's body_of does and is the
half that matters: a redefinition is one store, and it has to reach call sites
that already exist.
What is emitted, all of it behind dev:
- one cell per function in .data, .globl, initialised to the body this build
compiled. Spelled exactly as Emit.cellname spells it, because the point of
having one here is that an LLVM-built module binds
@"flan.cell.<n>" = external global ptr against it. nm -D over the two
builds of the same program gives identical sets of 68 cell symbols.
- the cell load placed after the arguments, which emit.ml has as a
load-bearing comment: a redefinition landing between two calls must not
land in the middle of one. CallPtr stays the other way round.
- the flan_dev_reg_enable constructor, which arms the allocation registry.
Not emitted: Emit.cellptr, the deeper spelling for a name the host was never
built with. It cannot arise in a whole-program build and belongs with the
redefinition module that would introduce one.
The --x86 --dev refusal is relaxed, and the argument is that flan dev never
reaches this fork: --x86 is read only by flan build, and the daemon builds
host and modules through Build.executable / Build.shared without it. So the
flag means a host whose call sites are redefinable, and nothing claims the
module that would redefine through them exists.
Two things were needed to believe any of that. First, the corpus with --dev on
both sides: 97 MATCH, 0 DIFFER, same as without it. Before the constructor was
added that read 96/1 — registry.flan asks (live? ...) and got four zeroes,
which is the whole of what a dev host does differently besides the cells.
Second, and the corpus cannot do this one: a dev build starts with every cell
pointing at the body this build compiled, so it prints what a release build
prints whether anything reads the cell or not. spike/x86/cells.sh preloads a
shared object whose constructor dlsyms flan.cell.twice and stores a different
body there -- the one store a redefinition ends in, done from outside, no
compiler involved. Both dev builds then print the new answer for a direct call
and for a function value, and both release builds are unchanged, which is what
says the change came from the indirection and not from symbol interposition.
One thing the later lane inherits, now written in both headers rather than
left to be discovered. x86.ml licenses its own calling convention on the
grounds that a dev build is compiled entirely here and a release build
entirely by LLVM, so the two never meet in one process. A cell an LLVM-built
module can store into is the first thing that could make that false: the
conventions agree on scalars and disagree on every aggregate, so an
Emit.redefinition module dlopened into an --x86 host would be right until the
first redefined function took or returned a struct. The answer is a
redefinition emitter here, not a classifier.
The transfer channel was the only thing between 41 programs and the
corpus. It is there now: a guard after every Flan call, a landing pad
per restart-case, handler-bind and with-allocator, a transfer exit per
function that runs its fdefers, and check_at and check_slice, which
could not exist until the guard did.
Measured by what the programs print and what they exit with, never by
reading bytes. spike/x86/survey.sh builds every program in
test/programs both ways and diffs stdout and the exit status; it did
not exist, so it is here too, and it is the progress meter.
before 41 MATCH 1 DIFFER 41 refused by name
after 83 MATCH 0 DIFFER 0 refused by name
The one DIFFER was bounds.flan, and it was the honest answer to
"--x86 is silently a --no-bounds-checks build". It is not one any
more: check_at and check_slice signal through the channel exactly as
emit.ml's do, so a bounds violation signals, a restart-case catches
it, and an unhandled one exits 134 on both backends. The transitional
refusal that would have said so retired before it was written.
check_no_transfer is not removed, it is narrowed to the one place the
argument still holds: a global's initialiser runs from
flan..init-globals, before main and before anything can handle
anything, so a transfer out of it has nowhere to go.
Four bugs, and three of them are the shape item 16 predicted -- code
that reads correctly and answers wrong, found by output and not by
objdump:
- The body fell through into the transfer exit, so every fdefer ran
twice on a normal return. emit.ml cannot have this bug: its ret
terminates the block.
- A Vec crossed to the runtime as the address of a *copy*, so pushes
grew the copy and an in-bounds (at v 1) signalled against a length
of zero.
- ucomis sets CF, ZF and PF together for a NaN, so sete answered true
for (= x x) and the prelude's NaN test never fired: (/ 0.0 0.0)
formatted as -9223372036854775808. Flan's comparisons are LLVM's
ordered ones, so < and <= swap and =, != take a setnp beside them.
- A union read field 0 through the struct table and was refused by
name rather than laid out as a tag and a payload.
And one that could not have been found later: emit_globals_init stored
a null *into* the channel slot rather than a cell address into it,
which is a null pointer for every callee to write through. Harmless
while nothing could transfer; a fault the first time a guard loaded
through it.
x86.ml was an encoder and a frame model with nothing calling it. It now
lowers a whole Tast.program to an assembly file, and `flan build --x86`
hands that file to the same clang invocation the LLVM path uses, against
the same runtime objects. The flag is off by default; LLVM stays the
release backend and the default one.
Three programs, built both ways and compared by what they print and what
they exit with rather than by reading bytes: exit 0; a dotimes that
prints; and a fizz over a call, an if, a remainder and two string
literals. All three agree with the LLVM build.
The measurement decided the target. hist.ml over the fizz program shows
no Signal, no Handled, no RestartCase — a loop that prints does not drag
conditions in. What does is the bounds check and the allocator, and
neither is in the reachable set of a program that prints a number.
That is why there is no transfer guard here, and check_no_transfer is
what makes the omission sound rather than hopeful: if nothing reachable
can write the channel, no call can return with it set. It is a
whole-program property, so it is checked once per build and the build
stops with the node's name when it fails.
Build.cachedir sat under TMPDIR, which dune makes private per run, so no
test run ever reused an object and every build in the suite was cold. It
moves to $XDG_CACHE_HOME/flan/objcache (FLAN_CACHE_DIR overrides), which
is safe because the keys are total: compile_c digests the source text,
the compiler's stamp and every flag; wasm_resource_dir digests the
builtins archive; compiler_object digests flan.cmxa and flan.a. Writes
were already .tmp-then-rename, so concurrent dune jobs are fine.
Macro.key was the one key that was not total -- prelude text plus the
call's forms, and nothing about the compiler whose codegen produced the
.so it names, which is dlopened straight back into this binary. Under a
per-run TMPDIR that never showed; under a durable cache it is a stale
expander that crashes rather than a compile error. It carries the
compiler's stamp now, handed across start_merged's exec in
FLAN_COMPILER_STAMP because a merged dev binary lives at a per-session
path and keying on that rebuilt a macro module every dev start.
Measured on dev-repl.flan, launch to bound socket: 2.0s cold against
0.48s warm. Whole-program flan build: 1.44s against 0.06s. Full dune
test 25.7s/30.1s before, 24.0s after, user CPU ~50s down to ~34s.
And the await: one timer covered two waits, a build then a bind, so
'the daemon never listened' was a wrong diagnosis of a build that had
not finished. listening now polls the process alongside the socket and
says which -- exited with a status, or still running and therefore still
building. A daemon that dies fails in milliseconds instead of costing
the whole timeout. Thirty seconds, down from a minute, because the build
it waits on is warm now.
Running a macro means compiling it and loading it into the compiler, and the
step that reads as small in NEXT.md is not: OCaml has no dlopen for ELF, and
lib/dune had no foreign_stubs. So the boundary is built first and the expander
not at all. lib/dynload_stubs.c is the whole of it — dlopen, dlsym, a
four-argument call into a macro thunk, and a peek/poke family, because OCaml
cannot address the raw memory a Form image has to be laid out in.
Nothing aggregate crosses to C. The unions lane verified a union's memory
layout against clang, which is a different claim from LLVM's convention for an
aggregate passed or returned by value in hand-written IR, so Emit.macro_thunk
wraps every macro in void(ptr,i64,ptr,ptr): the slice is built and the result
stored on the LLVM side, and the compiler's side is four pointers.
Build.macro_module links the runtime in rather than declaring it external, so
the module has no undefined symbols and the compiler's own link needs no
-rdynamic. That is the difference from Build.shared, whose host is a running
Flan program.
defunion Form and the list-building surface quasiquote will desugar into are in
the prelude. Form mirrors Form.value and not Form.t: no loc field, so the
compiler stamps the call site's location onto everything a macro returns.
The compiler builds. dune test was not run, and Form's layout is asserted
nowhere — NEXT.md's new handoff section says what the three numbers are, what
the next two commits should be, and the four decisions this made that the
design did not settle.
Three things stood between the flagship program and the web target, and each
is answered here rather than worked around.
The brush was a path. (rl/load-texture "brush.png") hands raylib a filename to
open, and a bare relative path means nothing on a target with no filesystem.
It is (embed "brush.png") now, decoded through a new binding —
LoadImageFromMemory, declared (Ptr u8) plus an explicit count because the shim
generator refuses a slice parameter and says so, with a Flan wrapper taking
the slice apart exactly as collision-point-poly? and load-font-ex already do.
One decode now serves both textures: the unflipped upload first, then
ImageFlipHorizontal in place, then the mirrored one. load-texture and
load-image lose their only call site in this repository; that is deliberate,
because a path-based load is the thing that cannot work here.
A package's C may now be addressed to one target, the way a link line already
could. A .c file may carry a tag before its extension — flan_agent.web.c — and
on that target it is compiled and *replaces* the untagged file of the same
base name. Replacement rather than plain tagging, so that teaching a package
about a new target is additive: the file that was right on three targets is
not renamed to say so. Selection is in Build and not in Load, for the reason
select_lflags gives.
The dev agent on the web is a no-op, and the reasoning is written at length in
vendor/agent/flan_agent.web.c. Short version: the agent is a socket server and
a browser has no sockets, so the missing <sys/time.h> was the surface and not
the cause. Refusing vendor:agent on a web target was the other candidate and
is ruled out by arithmetic — Flan has no conditional compilation, sand.flan
calls agent/start unconditionally, Reach cannot prune a package something
reachable calls into, so a refusal means the program does not build for the
browser at all. This does not contradict the `barf` decision made earlier
today. `barf` is asked to make something durable, and a no-op returns success
to a program that now believes bytes are on disk. The agent is asked to accept
redefinitions, and on the web there is no editor, no socket and no session —
--dev is refused by name on every wasm target — so there is nothing to lose.
sand.flan already says the same of a native release build at the call site.
test/test_web.ml builds sand.flan for the browser and reads the module for
brush.png's own bytes, whole. Not "IHDR": stb_image carries that string itself,
linked in from raylib, so it would pass on a build where the embed emitted
nothing. It is not run — node has no DOM, so main reaches InitWindow and dies
inside glfwInit on `window is not defined`, which says the module is live and
nothing about whether the canvas paints.
test/dune gains brush.png, because an embed is read by the checker and the
headless case reaches sand.flan through ../../ from a sandboxed _build.
test_session's C-c C-k case now passes ~origin, which is what both editor
paths already send; omitting it was testing a request nobody makes.
dune test is green. Docs follow in the next commit.
flan build --target=web produces a page, its JS and a .wasm. The two wasm
targets share the word and almost nothing else, so is_wasi and is_web are
separate predicates and is_wasm is their union — the union is exactly the
facts about the machine, 32-bit pointers and no dlopen, which is what the
refusals are about.
Everything the wasi target has to find by hand is what emcc already is: no
sysroot, no builtins archive, no shadow resource directory, and no
__main_argc_argv shim, because emscripten's start code calls main under that
name. target_flags for web is empty and the only thing checked is that emcc
exists. The one fact this rests on is that emcc takes a .ll on its command
line, so Emit's output needs no change.
The main loop is -sASYNCIFY rather than emscripten_set_main_loop, which
BUILT.md predicted. The prediction had the browser right and the cost wrong:
set_main_loop wants the loop body as a callback, so every example that writes
(until (rl/window-should-close?) ...) would be split by hand into an init and
a tick and would stop being the native program. raylib's web platform is built
for asyncify instead — WindowShouldClose on PLATFORM_WEB is an
emscripten_sleep(16) that returns false — so the loop yields at a call it
already makes and no example changed a character. Asyncify goes on every web
link, because whether a program blocks is not a question Build can answer and
a per-program flag set is a per-program cache key.
A link line may now be addressed to one target — @native, @wasi, @web — and
${NAME} expands from the environment. The selection is here and not in Load,
which reads the file, because Load resolves imports before a target is chosen.
The object cache now keys on whichever compiler the target uses, so an emcc
object and a clang one of the same source cannot collide. The refusals name
the target that was asked for; --sanitize on web says the weaker truth, that
emscripten ships an ASan and nothing here has ever run it.
ASan is an LLVM pass but instruments only functions carrying
sanitize_address, which clang's C frontend adds and nothing adds to IR
written by hand. Passing -fsanitize=address to the clang run over the
.ll therefore instruments flan_rt.c and not one instruction of Flan: an
out-of-bounds read of a defvar array, built --no-bounds-checks, printed
its garbage and exited 0. With Emit naming an attribute group on every
define, the same program reports global-buffer-overflow in flan.main.
UBSan has no such lever. Its checks are branches the C frontend emits to
__ubsan_handle_*, not a pass, so -fsanitize=undefined covers the runtime
and nothing else; (<< 1 32) still goes unremarked. Recorded where it
will be read rather than discovered again.
The flag does not force -O0 the way --debug does -- the UB worth finding
is what the optimiser does with it -- and it does pull in -g, since a
report with no line costs more than the build. compile_c's cache key now
digests the same cflags list the command line uses, because an
unsanitized flan_rt.o served out of the cache links fine and reports
nothing.
--debug is a third flag beside --dev and the optimisation level because it
answers a third question. --dev is "can I redefine this while it runs";
--debug is "can I stop it and read it". Either is useful without the other,
and a REPL session that is not being stepped should not pay for DWARF.
Not implied by -O0 in particular, for a reason already written down in this
file: the acceptance table runs the same programs at -O0 and -O2 to compare
the emitted IR against what mem2reg makes of it. If -O0 pulled in debug info,
every one of those comparisons would be against a different module.
It does imply -O0 downwards, and sets it. The whole mechanism is an
llvm.dbg.declare hanging off an alloca, and mem2reg deletes the alloca.
Refused for wasm32 by name. The member offsets in the DWARF are computed for
the host — ptr is 8 bytes — and wasm32's pointer is 4, so a slice's len sits
at byte 8 there and byte 16 here. Emitting the host numbers would hand a
debugger a confident wrong answer for every slice and every struct holding
one, which is the exact failure this project keeps meeting at the FFI
boundary. Silence would be worse than the refusal.
-g reaches the C compiles too, and joins compile_c's digest key with it, or
an object built without it would be served to a build that asked for it.
vendor/raylib has no C in it any more: shim.c is deleted and its 84 wrappers
are emitted from declare-c, which names the library's function in the library's
own signature. The reason the shim exists is unchanged - a small struct's
calling convention is a per-target classification and clang reproduces it for
free - but writing it by hand has stopped.
declare-c is a second form rather than a change to declare, because the two make
opposite claims about the same shape: (declare start-raw [path string] ...) says
the symbol takes ptr+len, and (declare-c init-window [... title string] ...)
says it takes a NUL-terminated char*. No structural rule separates them, so the
author says which.
The merge needed two fixes that neither lane could have found alone.
Load's uses-walker matches decl_kind exhaustively and did not know DeclareC, so
the reachability work and the generator did not compile together.
And the generated C is now emitted in parts keyed by the wrapper's own C symbol,
not as one translation unit. Reach.link drops the bindings nothing reachable
calls; a single TU holding every wrapper referenced every raylib symbol, so
sand-headless - which deliberately links no libraylib, and is the reason Reach
exists - failed at the link with undefined references to GetTime and its
neighbours. The first attempt keyed the parts by Flan name and broke the other
way, dropping a wrapper that was called: the flattened declaration is named
foo-c when a Flan wrapper is generated over it and foo when none is needed, so
the Flan name is not one thing. The wrapper's C symbol is what the declaration
binds in both branches.
Worth recording how close that came to passing: the acceptance suite died with
an exception rather than printing FAIL, so a grep for failures counted zero and
the suite looked green. Only the count of reporting suites - ten where there had
been eleven - showed it.
84 hand-written C wrappers is the shape of a job the compiler should be
doing. The reason the shim exists is unchanged and is not negotiable: a
small aggregate's calling convention is a per-target classification, not
part of its layout, and reproducing x86-64, arm64 and wasm32 inside
emit.ml is three classifiers to keep correct forever, where a mistake
reads as a field full of garbage rather than as a link error. clang does
it, per target, for free. So the C stays; the typing of it stops.
declare-c names the library's own function in the library's own
signature, and Shim emits the typedefs, the extern prototype, the
flattening wrapper and the flattened declaration the Flan side calls.
It is a second form rather than a change to declare because no
structural rule can separate them: (declare start-raw [path string] i32
"flan_agent_start") means the symbol takes ptr+len, and (declare-c
init-window [w i32 h i32 title string] "InitWindow") means it takes a
NUL-terminated char *. Same shape, opposite claims. declare is
untouched, so sqrtf and vendor/agent keep working unedited.
The generated C rides on Tast.program rather than beside it, so the CLI,
the REPL and the acceptance table all carry it without being told about
it. `flan shim` prints it, because a wrong binding is wrong in a wrapper
that is otherwise on no disk anywhere.
The wasi-sdk candidate had an LLVM version in it, which moves release to
release — so the path advertised as the proper article would have matched only
by coincidence, while the emscripten one beside it was derived. Both are
derived now.
calc-me on wasm32 covers what the other three cases cannot: flan_argv hands
Flan an array of flan_slice built in C, so what it pins is the element stride
of a ptr+len pair — 16 bytes native, 12 on wasm32 — rather than a field
offset. It is also the claim in this file's own header, that the table runs on
the second target, honoured for the first time.
flan emit refuses --target rather than stripping it. The IR really is
target-free, so ignoring it is correct and silence about it is not.
wasm32-wasi needs a sysroot clang does not know about and a builtins archive
Fedora does not ship, and both have to reach the C compiles as well as the
link — flan_rt.c includes <stdio.h> and never got past it. The flags are
computed once and the whole list, not just the triple, is in the object cache
key: repointing a sysroot must not be served a stale .o.
Fedora ships no wasm libclang_rt.builtins.a and clang's resource directory is
root-owned, so a shadow one is built under the object cache with the archive
under the name clang looks for. The archive substituted is emscripten's
libcompiler_rt.a, a different triple built by a different clang; wasi-sdk is
the proper article and the comment says so, because a session reading "wasm32
works" should know which joint is glued. Nothing found means a refusal naming
every path tried.
The entry point is the other thing no triple tells you: wasi-libc calls
__main_argc_argv, the .ll says @main, and the mismatch links clean and then
traps on a weak stub. Two lines of C bridge it, and the asm label in them is
why the shim is not an infinite self-call.
--dev and Build.shared are refused for the target rather than half-supported:
both are dlopen, which wasm32 has no equivalent of.
floor, ceil and round over f32, which is what a position and a tile coordinate
are here. The only rounding mode available is the cast's truncation toward
zero, so each of these is that cast plus the correction the mode does not
make, and the content is which inputs make the cast itself undefined. NaN
fails every comparison, so it needs its own (not (= x x)) and nothing else
finds it; the infinities fall out of the magnitude test; and above 2^23 an f32
has no fractional bits left, which makes returning the input there the exact
answer and also the guard that keeps the cast inside i32.
round is half away from zero, written as floor of the magnitude and mirrored.
The obvious (floor-f32 (+ x 0.5)) is wrong twice: half-up rather than
half-away, so -2.5 comes out -2, and at the largest f32 below 0.5 the addition
alone rounds to 1.0 and answers 1 for a number under a half. Both are in the
table, which is why every case there is a negative or a half.
sqrt is the decision in this commit and it goes out to libm, which is a change
to the release link and so is said out loud. Every other number in the prelude
is reachable from the four operations and a cast; a square root is not.
Newton's method needs a starting guess, the good guess comes from
reinterpreting the exponent bits, and the language has only value-preserving
casts - no bit-cast between f32 and u32. Without one the iteration needs a
scaling loop to normalise and still produces a result that is merely close,
which is the one thing a standard library must not hand back. IEEE-754 makes
sqrt correctly rounded, so libm's answer is the same bit pattern on native and
on wasm32; for this function the byte-identical argument points at C rather
than away from it.
The cost is -lm on every link, and its placement matters. It goes after the
objects, not in the leading flags, because --as-needed drops a library named
before the object that wants it. Worse, at -O2 LLVM folds most sqrtf calls
into the hardware instruction and the symbol never has to resolve - so this
looked linked before the flag existed and failed only at -O0, which is exactly
why the table runs both. Untested against --target=wasm32: wasi-libc ships
libm.a as a stub because the symbols live in libc, so it should be inert
there, but nothing here exercises it.
The better fix is not in this lane. llvm.sqrt.f32 as a builtin in check.ml and
emit.ml is one instruction, no symbol and no flag, and it belongs to whoever
owns the compiler.
vendor/agent/flan_agent.c calls flan_dev_result_get, which lives in flan_dev.c,
which build.ml compiled only for a dev build - so flan build sand.flan died at
the link with an undefined symbol. A regression from 7ce1d09, where C-x C-e
gave the agent a result to report.
A package's C sources are collected whatever main does, so the agent's C is in
every build that imports it. flan_dev.c is now compiled into all of them.
Nothing in a release build reaches it: the compiler emits a registry lookup
only for a name the host was not built with, and without cells there is no such
name. The table is BSS, so the cost is address space rather than binary size,
and -rdynamic and the cells are still what --dev means.
test_agent.ml now links the same program both ways. It runs only the dev one -
with no cells the agent refuses every module, so linking is the whole claim.
Editing a defvar or a defn is a symbol the host exports. Adding one is not:
there is nothing to bind to and ELF cannot grow a symbol. runtime/flan_dev.c is
the two lookups that cover it - flan_dev_cell for a new function's cell,
flan_dev_global for a new global's storage - both idempotent, so the second
module to mention a name gets what the first one got. That is the whole point:
two modules with their own copy of a new function would each call their own,
and redefining it would update one of them.
The compiler picks per name. A name the host has is a symbol and costs one load
at a call site; a name it lacks is a registry lookup cached at install time in
a module-local slot, and costs two. The common case pays nothing for the
general one.
The redefinition unit is now a list of top-level forms rather than one
function. It has to be: v3 of the fixture adds a var and uses it from a
redefined bump, and splitting that into two loads leaves a module referring to
storage that does not exist yet. C-c C-c passes one name, C-c C-k passes a
file's worth, one path either way.
Four rules, each silent if broken. Every lookup resolves before any body is
published, or a caller reaches a function whose slots are still null - asserted
on the emitted flan_reload_install, since it cannot be race-tested.
flan_dev_global refuses a size change, which is the layout-drift rule's first
enforcement point rather than another exception to it. Nothing is ever
dlclosed, because a cell holds an address inside a module's text. And the table
is fixed capacity, because a module holds a cell's address for as long as it is
loaded and a realloc would strand it.
The test that separates this from a plausible wrong version is v4, which
redefines a name v3 introduced at run time. v3's bump is already installed and
is not rebuilt, so it picks v4 up only if its call goes through a cell both
modules found by the same name. Had v3 cached the function's address instead,
every other assertion would still pass and the transcript would read 246
instead of 432.
Sizes are spelled LLVM's way, ptrtoint getelementptr null 1, rather than by a
layout calculator in OCaml that would have to agree with LLVM's on every
target.
Two things, and either alone is useless, so they are one commit.
Emit.redefinition compiles one function into its own module against a host
that is already running. What it does *not* define is the design: a global is
external, so state survives a reload and sand's grid is not reset by editing
the code; every other function is a declare, so a redefined settle calls the
host's move-grain rather than a frozen copy; there is no main. Build.shared
puts that text through llc + ld -shared. ld, not clang, because a shared object
is allowed undefined symbols and that is the whole mechanism - and because the
driver is 50ms of a 20ms job. Measured here: llc 16ms, ld 3ms, dlopen 0.04ms.
Loading a body is not installing it, though. A call bound at link time cannot
notice a new one, so a dev build routes every Flan-to-Flan call through a cell
- a mutable global holding the address of the function that is current - and a
module publishes itself with one store. The cell load is emitted after the
arguments, so a redefinition between two calls cannot land inside one.
Three details that are not free choices. flan_reload_install is a named
function rather than an ELF constructor, because the agent has to choose when
the store happens and a constructor would do it during dlopen, mid-frame, on
whatever thread called it. A redefinition's own body is hidden, because default
visibility in a shared object is interposable and that applies to taking the
address too: plain @"flan.bump" inside the module resolves to the host's copy,
so the installer would publish the function it was replacing and the reload
would silently do nothing. And -rdynamic is what exports the cells at all, so
it and cells are one flag: Build.opts.dev, flan build --dev, the first time
opts means something semantic rather than an optimisation level.
The test is one process, because two runs would prove nothing about a swap,
and two .so paths, because dlopen caches by path and would hand back the first
handle. Every call in it goes through outer, compiled once into the host and
never rebuilt, so a changed answer can only mean its call site followed. v2
recurses through its own cell, which is the interposition case; it would print
the old body's text if it did not. helper differs between the fixtures purely
as a tripwire for a module that grew its own copy.
LLVM cannot fold the indirection - the cell is an external mutable global - and
a --dev calc-me keeps 46 indirect calls at -O2. values, machine and
sand-headless now run as dev builds in the acceptance table too; the sand hash
is the one result that would notice a call reaching the wrong function.
