* libguile/memoize.h:
* libguile/memoize.c (MAKMEMO_CALL_WITH_PROMPT, memoize, unmemoize):
Remove the @prompt memoizer in favor of recognizing call-with-prompt
primcalls. Rename SCM_M_PROMPT to SCM_M_CALL_WITH_PROMPT, and pass a
thunk instead of an expression so that it has normal applicative
order.
* libguile/expand.c (PRIMITIVE_REF, PRIMCALL, expand): Produce primcalls
from forms whose car is a primitive.
(expand_atat): Recognize (@@ primitive FOO) as being a primitive-ref.
* module/ice-9/boot-9.scm (call-with-prompt): Instead of dispatching to
the wonky @prompt memoizer, residualize a primcall to
call-with-prompt. The memoizer will DTRT to allow call-with-prompt to
be interpreted correctly without needing an additional binding.
* module/ice-9/eval.scm (primitive-eval): Change the 'prompt clause to a
call to call-with-prompt.
* module/language/tree-il/primitives.scm: No more need to recognize
@prompt.
* libguile/eval.c (eval): Adapt to SCM_M_PROMPT renaming to
SCM_M_CALL_WITH_PROMPT, and apply the thunk.
* libguile/throw.c (pre_init_throw): Adapt to scm_abort_to_prompt_star
rename.
* module/language/tree-il.scm (<tree-il>): Remove pre and post fields
from <dynwind>. A dynwind now assumes that in normal entry and exit,
that the code runs the winders and unwinders using <seq> and
<let-values> and such things.
(parse-tree-il, unparse-tree-il, make-tree-il-folder, pre-post-order):
Adapt <dynwind> users.
* module/language/tree-il/analyze.scm (analyze-lexicals):
* module/language/tree-il/compile-glil.scm (flatten-lambda-case):
* module/language/tree-il/cse.scm (cse):
* module/language/tree-il/debug.scm (verify-tree-il):
* module/language/tree-il/effects.scm (make-effects-analyzer): Adapt.
* module/language/tree-il/peval.scm (peval):
* module/language/tree-il/primitives.scm (*primitive-expand-table*):
Produce tree-il that calls the winder and unwinder. Recognize
singly-valued dynamic-wind expressions.
* test-suite/tests/peval.test ("partial evaluation"): Add tests.
* module/language/tree-il/peval.scm (singly-valued-expression?): Add
support for conditionals. In the future we should add more
expressions here.
(peval): Don't inline values into the body of a dynwind, as that could
cause the consumer to run in the wrong dynamic context.
If the producer is singly-valued and the consumer just has a rest arg,
reduce to "let" and cons up a list in the consumer. This may reduce
further.
* test-suite/tests/peval.test ("partial evaluation"): Add a test.
* module/language/tree-il.scm (tree-il-fold): Implement using
make-tree-il-folder. This is an incompatible change: there is no more
"leaf" procedure, and tree-il-fold only works on tree-il and not
lists.
* module/language/tree-il/analyze.scm (<tree-analysis>, analyze-tree):
Adapt to tree-il-fold change, losing the "leaf" handler.
(unused-variable-analysis, unused-toplevel-analysis)
(unbound-variable-analysis, arity-analysis): Adapt to tree-analysis
change.
* module/language/tree-il/canonicalize.scm (tree-il-any)
* module/language/tree-il/cse.scm (build-assigned-var-table)
* module/language/tree-il/peval.scm (tree-il-any, build-var-table)
(peval): Adapt to tree-il-fold change.
* test-suite/tests/tree-il.test ("tree-il-fold"): Adapt tests for new
interface and expectations.
* module/language/tree-il.scm (pre-order): Re-implement in terms of
pre-post-order, and rename from pre-order!.
* module/language/tree-il/primitives.scm (expand-primitives): Adapt to
pre-order change, and rename from expand-primitives!.
* module/language/tree-il/optimize.scm (optimize): Adapt to
expand-primitives! change, and rename from optimize!.
* module/language/tree-il/compile-glil.scm:
* module/system/repl/common.scm:
* test-suite/tests/cse.test:
* test-suite/tests/peval.test:
* test-suite/tests/tree-il.test: Adapt to expand-primitives and optimize
changes.
* module/language/tree-il.scm (pre-post-order): New helper, like
pre-order! and post-order! but not destructive.
(post-order): Implement in terms of pre-post-order, and rename from
post-order!.
* module/ice-9/compile-psyntax.scm (squeeze-tree-il):
* module/language/tree-il/canonicalize.scm (canonicalize):
* module/language/tree-il/fix-letrec.scm (fix-letrec):
* module/language/tree-il/primitives.scm (resolve-primitives): Use
post-order, and rename from the destructive
variants (squeeze-tree-il!, canonicalize!, etc). Adapt callers.
* test-suite/tests/tree-il.test (strip-source): Adapt to post-order.
* test-suite/tests/cse.test:
* test-suite/tests/peval.test:
* module/language/tree-il/optimize.scm: Adapt callers.
* module/language/tree-il/peval.scm (peval): let-values with a consumer
that has only one argument is the same as let.
* test-suite/tests/peval.test ("partial evaluation"): Add test.
* module/language/tree-il/compile-glil.scm (flatten-lambda-case):
Support calls and tail-calls with more than 255 formals.
* test-suite/tests/tree-il.test ("many args"): Add a test.
* module/language/tree-il/peval.scm (peval): Add a special-case inlining
pattern for apply to a let-bound rest arg that preserves effect
ordering.
* test-suite/tests/peval.test ("partial evaluation"): Add a test, and
update an older test with a better result.
* module/language/tree-il/peval.scm (peval): Correct comment on
find-definition, and allow a find-definition to fall back on a source
expression. Avoid copying non-constant expressions.
* test-suite/tests/peval.test ("partial evaluation"): Add a test that
inlining rest arguments works with complicated argument expressions,
and a test that order of effects in rest args is preserved.
* module/language/tree-il/peval.scm (peval): Move up the find-definition
helper. Use it to speculatively destructure conses and lists into the
tail position of an `apply' form.
* test-suite/tests/peval.test ("partial evaluation"): Add tests.
* module/language/tree-il/analyze.scm (format-string-argument-count):
Handle ~t and ~k options case-insensitively.
~! ~| ~/ ~q and ~Q should not update the min-count or max-count.
~^ returns the min-count and 'any
+ - # and ' should not increase the argument count.
* test-suite/tests/tree-il.test (*): Tests for new parameters.
* module/system/base/language.scm (<language>): Add 'for-humans?'
field, and export new 'language-for-humans?' predicate.
* doc/ref/compiler.texi (Compiler Tower): Document.
* module/language/assembly/spec.scm, module/language/bytecode/spec.scm,
module/language/glil/spec.scm, module/language/objcode/spec.scm,
module/language/tree-il/spec.scm, module/language/value/spec.scm:
Mark these languages as 'for-humans?'.
* module/language/tree-il/peval.scm (peval): Gensyms use whitespace as
an infix, not a dash. Helps gensym?-like procedures like those in the
unused lexical analysis.
Fixes <http://bugs.gnu.org/12883>.
* module/language/tree-il/cse.scm (cse)[find-dominating-lexical]: Fix
computation of the last argument passed to `unroll'.
Patch by Stefan Israelsson Tampe <stefan.itampe@gmail.com>.
* test-suite/tests/cse.test ("cse")["http://bugs.gnu.org/12883"]: New
test.
Moved scm_i_struct_hash from struct.c to hash.c and made it static.
The port's alist is now a field of 'scm_t_port'.
Conflicts:
libguile/arrays.c
libguile/hash.c
libguile/ports.c
libguile/print.h
libguile/read.c
* module/language/tree-il/primitives.scm (maybe-simplify-to-eq): New
helper procedure shared by expanders for 'eqv?' and 'equal?'.
(*primitive-expand-table*): Add expansion rule for 'eqv?'.
* test-suite/tests/tree-il.test ("primitives"): Add tests.
* module/language/tree-il/primitives.scm (*primitive-expand-table*):
Don't simplify 'equal?' to 'not' or 'null?', but only to 'eq?'.
* test-suite/tests/tree-il.test ("primitives"): Adjust tests.
* module/language/tree-il/primitives.scm (*primitive-expand-table*):
Refine test for fixnums to verify that the small integer is exact.
* test-suite/tests/tree-il.test ("primitives"): Add test.
* module/language/tree-il/primitives.scm (*primitive-expand-table*): Add
expansion rules for `equal?', when called with one constant and
another argument.
* test-suite/tests/tree-il.test (pass-if-primitives-resolved): New
macro.
("primitives"): New test prefix.
* module/language/tree-il/fix-letrec.scm: Use effects.scm for effects
analysis, instead of primitives.scm.
(simple-expression?, partition-vars): Adapt.
* module/language/tree-il/peval.scm (<operand>): Instead of having a
`residualize?' field, have it be a use count.
(peval): Adapt to <operand> change. Add function to kill uses of an
operand. Use it in the <prompt> inliner. Add another kind of
degenerate prompt to elide. We should really switch to CPS though, as
that will allow us to contify more aggressively.
* test-suite/tests/peval.test ("partial evaluation"): Adapt (while #t
#t) test, which was sensitive to how far the recursive inlining got.
Add a test for the degenerate prompt elision.
* module/language/tree-il/effects.scm (make-effects-analyzer): Be more
precise regarding the effects of the so-called
effect+exception-free-primitives: now we check their arities.
* test-suite/tests/cse.test ("cse"): Add a test that we don't
elide (cons 1 2 3) in effect context.
* module/language/tree-il/peval.scm (peval): Inline applications where
we know the contents of the tail.
* test-suite/tests/peval.test ("partial evaluation"): Add tests.
* module/language/tree-il/primitives.scm (*interesting-primitive-names*):
(*effect-free-primitives*): Add string->number, number->string,
integer->char, and char->integer to allow for constant folding and
better effects analysis.
* module/language/tree-il/cse.scm (cse): When trying to fold
conditionals, only look at entries in the database that were added in
test context.
* test-suite/tests/cse.test ("cse"): Add a test case.
* module/language/tree-il/analyze.scm (validate-arity): Fix for
applicable structs. Applicable structs are procedures, but not every
struct has a first slot, and not every struct with a procedure in its
first slot is applicable. Besides, the approach in this patch gives
better errors.
* module/language/tree-il/effects.scm (make-effects-analyzer): The
analyzer will take an optional second argument, a lookup procedure of
type sym -> exp. This can let the analyzer dig into calls to
lexically bound procedures.
* module/language/tree-il/cse.scm (singly-valued-expression?, cse):
Allow CSE to propagate lexicals to tail positions, if the expression
is singly-valued.
* module/language/tree-il/peval.scm (peval): If we can lift one common
test, see if we can lift others as well.
* test-suite/tests/peval.test: Add a test.
* module/language/tree-il/peval.scm (peval): Optimize common tests in
chains of "if" expressions, like those generated by matchers.
* test-suite/tests/peval.test ("partial evaluation"): Add a test.
* module/language/tree-il/analyze.scm (proc-ref?)[special?]: New
procedure.
Return #t for any toplevel-ref of `_'.
* test-suite/tests/tree-il.test ("warnings")["format"]("non-literal
format string using gettext as top-level _"): New test.
