260 Commits

Author SHA1 Message Date
53880eb685 A class slot's type names its own package's class first, an int widens into a float slot exactly when it round-trips, and a constructor's refusal names the call that was wrong 2026-09-25 13:29:30 +07:00
9aeee9d676 Merge branch 'master' into worktree-agent-a6ffe579d55d1892b 2026-09-25 13:16:58 +07:00
e7b01ea01c Merge branch 'master' into worktree-agent-adcfc27e0ceac8df4 2026-09-25 12:55:56 +07:00
9ba99d7d5b A dev build puts every closure handed to a named call on the collector's heap, since a redefinition can make the callee keep it 2026-09-25 12:55:56 +07:00
16427f8d07 A class slot may be a class or an Option, an int widens into a float slot when exact, a gained typed slot starts at its zero, and a migration never re-enters itself 2026-09-25 12:55:25 +07:00
1d219b526c Merge master into the typed class slots lane 2026-09-25 12:14:42 +07:00
c6cb018b9f Only a closure that outlives its frame takes a collector environment, and the collector reads a Vec's elements only through a block it knows is live 2026-09-25 12:12:25 +07:00
f629123561 A lowercase vec, ptr, option or map over types in a return slot is a near miss for its constructor, and a bare && is written back as the (and) that compiles 2026-09-25 11:56:56 +07:00
3892f1ed34 Merge master 2026-09-25 11:50:29 +07:00
d6217830ff A defn whose body stands in its return slot is told it has no return type, a misspelled type constructor is a near miss, and not=, ==, &&, || and their kin name the Flan builtin that computes the same thing 2026-09-25 11:50:23 +07:00
2682214499 A defclass slot may declare a type that every store into it checks, and set writes a declared slot 2026-09-25 11:40:08 +07:00
5062052425 Raylib's strings, models, file lists and matrix stack are bound, and a dev build can name what is still loaded at exit 2026-09-25 11:28:02 +07:00
05d86b243f A library resource is known by its identifying field, a Get or a consuming Load is counted as raylib means it, bare pointers are not resources, the report waits for FLAN_DEV_LEAKS, and a release build emits the code it did before the tracker 2026-09-25 11:27:41 +07:00
9e26275f9b A quasiquote nests as SBCL's does, a match takes an enum, and until is a prelude macro 2026-09-25 11:27:14 +07:00
494c57d764 Merge branch 'master' into worktree-agent-ae2997a1e90aa9e02 2026-09-25 11:18:12 +07:00
840b3b215e A Map's churn test fails a table that doubles instead of sweeping, and nothing recalled is cited as read 2026-09-25 11:18:12 +07:00
389e93ff09 A capturing fn's environment belongs to the collector, so a closure may outlive the frame that made it 2026-09-25 11:13:57 +07:00
d03b02b347 Merge master 2026-09-25 11:12:30 +07:00
0d771494a6 Merge branch 'master' into worktree-agent-a8f5ad274482eeb9c 2026-09-25 11:07:02 +07:00
8bebff61eb Merge branch 'master' into worktree-agent-a8f88805fa4af49cf 2026-09-25 11:01:36 +07:00
3f30b82bcb A return computes its value before its defers, and a dozen small refusals and constants say what the program meant 2026-09-25 10:54:09 +07:00
2f3544e7f5 Code after a return emits nothing, a wide element names the u64 array, and $ is refused in fn, handler, :keys and & binders 2026-09-25 10:53:52 +07:00
0a6ea7c32b A caller excused for going stale takes back the generic copies its failed check cached, and a package's stale callers are named at their own file and line 2026-09-25 10:51:44 +07:00
fe20f68ada The runtime's races are driven by tests, a destroyed arena's record is reused, and a dyn index names its site 2026-09-25 10:51:08 +07:00
a49ed2243a Merge branch 'master' into worktree-agent-a4be641f87966b149 2026-09-25 10:48:05 +07:00
0863b62711 A string returned from C is copied into the context allocator, Model, Mesh and FilePathList cross the raylib boundary, rlgl's matrix stack is a package, and a dev build reports the library resources a program never released 2026-09-25 10:44:49 +07:00
39d35f51db A function whose signature changed installs, and a caller compiled against the old one stops on StaleCall at the call
A dev cell carries its body's signature word beside the body, every call through a cell (and every function value taken from one) compares it with the word the site was compiled for, and the session lists the stale callers by file and line on the reply. Both backends, both installers; release builds have neither the word nor the compare.
2026-09-25 10:40:18 +07:00
e9e151e1a4 A match over an enum names its members as keywords and is refused when it misses one 2026-09-25 10:33:22 +07:00
83b244475a A refusal that suggests a fix suggests one that compiles, for vec-new, map-new and a near miss that is a value 2026-09-25 10:30:52 +07:00
6c9635cd53 A wide literal is refused for its range as an enum member, names the u64 cast where a dyn is wanted, and comes back wide from a macro 2026-09-25 10:26:03 +07:00
5581bafacb A dyn at, set-at or push that traps names the file, line and column it was written at 2026-09-25 10:25:28 +07:00
dd160aa31b A return computes its value before it runs its defers 2026-09-25 10:19:09 +07:00
e511174d3c No top-level value in the compiler goes unreferenced 2026-09-25 10:16:25 +07:00
d006a8b480 An array literal's first element types the rest, constant arithmetic folds at a bounded type variable, and a pointer or a plain union takes a byte fill 2026-09-25 10:14:24 +07:00
4a292b5a73 f64-inf, f64-nan, f32-inf and f32-nan are constants the compiler supplies 2026-09-25 10:07:22 +07:00
11dc230974 A defn- is private to its module, and the module's own macros may call it 2026-09-25 08:57:20 +07:00
793b0eea9e A defn- is private to its module, the package's own macros may write calls to it, and the refusal names the fix 2026-09-25 08:55:37 +07:00
00f116ea29 Six small refusals and conversions say what the program wrote 2026-09-25 08:41:31 +07:00
ff744120fc A release build's watch is only its value, next-error lets go of a stack that has resumed, and the frame a stop is in is always shown 2026-09-25 08:36:53 +07:00
cb0be62087 An integer written at or above 2^63 is a u64 and nothing else, and vec-new reads a type from its argument only when the builtin is what was called 2026-09-25 07:43:24 +07:00
6fd456c0a2 A function declared with defn- is callable only from the files of its own package 2026-09-25 07:42:40 +07:00
85b9f35a66 (watch "name" v) renders any value into the watch table the way print renders it 2026-09-25 07:22:59 +07:00
f142cfaadf A u64 constant above 2^63 can be written in decimal, and a cast's integer literal too wide for i32 is read at the cast's type 2026-09-25 07:16:42 +07:00
aa5052e401 A dyn value an expression answers is the reply's value, not the program's output 2026-09-25 07:14:13 +07:00
aa629da2e5 vec-new and map-new take a type expression where they take a type, so (vec-new [u8]) makes a Vec of byte slices 2026-09-25 07:10:48 +07:00
b4e19bebf8 A loop's initialisers are checked in order and each sees the names bound before it, as a let's do 2026-09-25 07:06:17 +07:00
45cc4d20e8 One file holds what five did
# Conflicts:
#	FIX.org
2026-09-21 21:07:40 +07:00
57fe91f303 Five records become one, and every citation lands somewhere
FIX.org, NEXT.md, DISCUSS.org, docs/DISCUSS.md and the session handoff at the
root are one TODO.org now: 293 entries under seven subsystem headings, each
carrying an org keyword that says where it stands. A DONE entry is a few lines
saying what was decided and what that rules out; the reasoning that would not
compress — the embedding spike and the four reports the hand-written x86
backend was built from — moved into docs/BUILT.md instead, and its entries
point there in one line.

Every entry was checked against the tree before it got a keyword, and the
prose was wrong in both directions. Things the deleted files called open were
built: the first-evaluation stall, main being redefinable, macro parameter
lists, the type-limit constants, the array constructors, the byte fills,
inc/dec, the discard's fontification, the Emacs buffers, rt_die's _exit, the
backtrace surface, and the acceptance failure that could print and still exit
zero. Things they called done were not: the backend reports' no-plan buckets
had gone stale in the other direction, the value-dependent defvar was
superseded rather than built, and macro-expansion source locations are on an
unmerged lane, so that entry is NEXT and names the branch.

Every comment that cited one of the five by name now cites a heading that
exists, in TODO.org or in docs/BUILT.md. The session reports under
docs/handoffs/ keep naming the files they worked on, because rewriting them
would falsify what those sessions did; each carries a note saying where the
content went.
2026-09-21 21:05:48 +07:00
Joseph Ferano
a14a872a44 A form a macro splices through is reported on the line it was written on 2026-09-21 19:16:45 +07:00
0c711942f7 A thunk named for the order it was minted in is a different signature after a reorder
Three from a review of the commit before this one, and the first is a dev
loop defect.  Naming the widening thunk thick/<count so far> put traversal
order into a symbol, and Session.compatible compares a reload's functions
against the running program's by name -- so swapping two calls in a body
renamed nothing the programmer can see and reported "thick/0 changes
signature, from [i32] i32 to [i64] i64.  Restart to change it."  The name is
the signature again, written by thick_enc: every type self-delimiting, an
atom as its length and then its spelling, so the boundaries mangle_ty loses
are in the string rather than inferred from the separators.  mangle_ty is
untouched -- it is load-bearing for the instantiation names a backtrace, a
Reach edge and a dev cell show -- and the memo stays keyed on the types.
fn-thunk-reload.flan is the pair of widenings and test_session reorders them.

The second was a silent miscompile.  bind_ty's Fn-pattern-against-CFn-argument
arm was structural, so it matched nested function positions too, and the
catch-up pass that builds the thunk compares the whole substituted parameter
list -- false when the mismatch is inside one, and the fallthrough handed the
argument over unchanged: one word where the instance declares two, 255 on
both backends with nothing said anywhere.  The widening is a value the caller
builds around the whole argument and there is nowhere inside one to build it,
so the arm is admitted at the top of an argument's type and nowhere else, and
the catch-up arm is total over the pair rather than guarded -- bind_ty's
fallthrough is Types.fits, which admits Never, so a binding can still succeed
over a pair the two words cannot bridge.  fn-generic-nested.flan is the
parameter position and fn-generic-nested-return.flan the return one.

And the third closes the hole the escaping inversion was for.  A match arm
binds the case's fields to slots and the store that fills them is inside the
branch, not in any form escape_check reads as a binding, so (match o (Some f)
f ...) handed back through a name what a case-field read cannot hand back at
all -- sound only because Some_ denies suspects.  An arm's slots of function
type are suspect now, for the same reason the read itself is.
2026-09-21 15:44:28 +07:00