TODO.org records what this lane settled and the one decision it left

TODO.org

@@ -73,11 +73,13 @@ is wrong where it was written rather than aborting the compile with no location.
 The prelude's own =unless= has not been converted and still answers a bare
 undefined name.
 
-** TODO gensym's counter restarts in a second module
+** DONE gensym's counter restarts in a second module
-The counter lives in the loaded module and a module is dlopened once per compiler
+CLOSED: [2026-09-25]
-process, so it is process-wide in practice — but the rounds already build more
+The counter is C data in the runtime (=flan_gensym_n=), and =lib/macro.ml= writes
-than one module for a program whose macros call macros. Seed it from the module's
+the compiler's own count into the module before every macro call and reads it
-index.
+back after. It counts across every module a compiler process loads — each round,
+the program's module, and every expansion in a session. Rules out a counter per
+module, seeded or not.
 
 ** TODO A quasiquote inside a quasiquote is refused
 Nothing counts nesting levels — not the reader, deliberately, and not the
@@ -141,11 +143,14 @@ as words the reader will not read back. =(/ 1.0 0.0)= is the only route to an
 infinity, and the constant folder is integers only, so it cannot be a =defconst=.
 Closing it needs a reader literal or a float-capable folding pass.
 
-** TODO A u64 constant above 2^63 cannot be written in decimal
+** DONE A u64 constant above 2^63 cannot be written in decimal
-The reader reads a decimal integer literal as a signed 64-bit number; hex is read
+CLOSED: [2026-09-25]
-as a bit pattern and works. The same limit has a second face: a cast's argument is
+A decimal between 2^63 and 2^64 is read as its bit pattern, as hex is. A cast's
-checked against the default type, so =(u64 2935910691)= is refused for not fitting
+integer literal that does not fit =i32= is checked at the cast's type; one that
-in an =i32=.
+fits keeps the =i32= default, so =(u32 -1)= still means what it did. It inherits
+hex's hole: a decimal above 2^63 at =i64= is accepted as a negative, and a refusal
+at a narrower type prints the negative number. Rules out a separate unsigned
+literal in =Form=, whose layout the prelude's =Form= mirrors.
 
 ** DONE {.row .col} binds same-named locals
 CLOSED: [2026-09-20]
@@ -790,14 +795,11 @@ program that does not type-check. Moot for anything that compiles; only the
 daemon's half-typed recompiles could feel it. A cheaper retry was tried and
 shelved because it changes which literal gets the nicer message.
 
-** TODO and's last operand gets a misdirected caret
+** DONE and's last operand gets a misdirected caret
-=(println (and true true (vec-new i32)))= puts the caret on the second =true=. The
+CLOSED: [2026-09-25]
-last operand of an =and= is the then arm and the then arm is typed first, so the
+Already fixed by 3672da2, which blames the arm that is not a compiler temp; the
-mismatch is blamed on the else arm, which carries the previous operand's location.
+caret is on the last operand and =test/test_flan.ml= asserts its column. Rules
-The fix is preferring the arm that is not a compiler temp when deciding whom to
+out relabelling the else arm, a bool sentinel, and inverting the condition.
-blame. Three others were considered and rejected: relabelling the else arm reads
-backwards, a bool sentinel reverts the =or= fix, and inverting the condition costs
-a =not= per operand.
 
 ** TODO Signature pairing's cold-rebuild edge
 Whether a parameter vector reads as one annotated parameter or two dyn ones
@@ -841,14 +843,20 @@ Iteration is built; the remaining refusal is generics. A =defn= has to name its
 types and =(defn map-keys [m (Map K V)] (Vec K))= has no =K=. The loop is three
 lines at the call site, where =K= is known.
 
-** TODO (vec-new [u8]) is refused
+** DONE (vec-new [u8]) is refused
-The element type must be a bare symbol naming a type, so a =(Vec [u8])= can only
+CLOSED: [2026-09-25]
-be made where the context names it. The fix is letting it take a type expression —
+The type positions of =vec-new= and =map-new= take a type expression: brackets, or
-the same parser that already reads =[u8]= in a parameter list.
+a parenthesised =Ptr=, =Option=, =Vec=, =Map=, =Fn= or =CFn=. Parse reads it with
+=texpr= into =Ast.TypeArg=; a bare name is still left for the checker to tell a
+type from an allocator. Rules out a type expression anywhere else in expression
+position.
 
 ** TODO An array literal cannot say it is [f32]
 A float literal defaults to =f64=, an array literal has no context, and a =let=
 has no annotation. Same shape as =(vec-new [u8])= and probably the same fix.
+Not the same fix: a bracket literal has no argument to put a type in. Decision:
+how a literal names its element type — a spelling of its own, or a =let=
+annotation.
 
 ** TODO A let binding takes no type annotation
 Everything under the surface is there — the binding carries a type slot and the
@@ -1808,12 +1816,12 @@ instrumented copy; the equivalent here is a dev-build-only instrumented
 redefinition, which the cell indirection already makes deliverable. Open:
 whether stepping suspends the frame loop, and what it does to a game's clock.
 
-** TODO A NaN cast says "does not fit", which reads as too big
+** DONE A NaN cast says "does not fit", which reads as too big
-=runtime/flan_rt.c:1008= covers every out-of-range float with one sentence, so
+CLOSED: [2026-09-25]
-=(i32 nan)= reports the =i32= bounds as if the value had overshot them. NaN and
+Two more =ArithError= codes: 5 for a cast of NaN and 6 for a cast of an infinity,
-the infinities convert to no integer at all and want saying so by name. Found
+each with its own sentence. Both backends choose the code on the cold path, so the
-by filling a struct holding an =f32= with =(filled 0xFF)=, where every bit set
+guard is still two compares. =lhs= and =rhs= still carry the range. Rules out
-is NaN.
+carrying the float value in the condition.
 
 ** TODO The break buffer prints fields, not the sentence the runtime wrote
 =ArithError — op 4, lhs -2147483648, rhs 2147483647= where the runtime's own
@@ -1837,12 +1845,12 @@ It prints the source line and carets for the stop (=flan-cnr.el:197=) and lists
 frames, but no key opens the file at that line. Wants RET-on-a-frame, or =M-.=,
 and =next-error= over the frame list.
 
-** TODO loop's bindings should be sequential, like let's
+** DONE loop's bindings should be sequential, like let's
-=check_loop= (=lib/check.ml:4741=) checks every initialiser before binding any,
+CLOSED: [2026-09-25]
-so =(loop [curr-r r  next-r (inc curr-r)] ...)= cannot see =curr-r= and the
+=check_loop= binds each name before checking the next initialiser; =recur= still
-refusal reads as an unknown name. Every binding form is sequential — there is
+rebinds all at once. No other form had the gap: =let= was already sequential,
-no =let*= here and there is not going to be one. Check the other binding forms
+=dotimes= binds one name, and =fn=, =defn=, =match= and the handler and restart
-for the same gap while fixing it.
+clauses bind parameters with no initialisers.
 
 ** TODO C-c C-c reports one error, not every error in the form
 Whole-file paths use =Check.program_all= and report every bad declaration. The