Namely, not the singular record signature, but the signature
constructors of `define-record-procedures-parametric'.
(We'd done this before, but that patch got reverted, without the
doc part getting reinstated.)
Such as this, submitted by Torsten Grust:
(define-record-procedures-parametric stream_ stream-of
make-stream
stream?
(stream-head stream-tail))
(define stream
(lambda (t)
(signature (stream-of t (-> (stream t))))))
(: from (number -> (stream number)))
(define from
(lambda (n)
(make-stream n (lambda () (from (+ n 1))))))
(: foo (stream number))
(define foo (from 1))
The problem was that the arbitaries got evaluated too eagerly.
the state that online check syntax uses (indirectly via scribble)
to lookup documentation.
since the kill-safe abstraction moved some computation over to
another thread, the values of various parameters (current-directory
in this case) were not preserved. So when scribble then ended up
resolving a module path with the wrong current directory, this
bad value got cached. Then later, when check syntax tries to figure
out what file to open to jump to it, it got the cached bogus value
(even though current-directory was right when check syntax asked
for the filename).
closes PR 12538
Rename `read-intern-literal' to `datum-intern-literal'.
Interning is needed only in `read-syntax' or `datum->syntax' to
set up the invariants that the bytecode compiler needs for cross-module
optimization. When `read'ing numbers from a data file, meanwhile,
interning slows things down a lot and doesn't seem worthwhile.
Pass a pointer to the thread-local table on entry to JIT-generated
code, instead of having the JIT-generated code call a C function
to get the table. This doesn't seem to improve performance on my
machine, but it generates less code and is probably faster in
some cases.
