This allows it to actually use all CPU cores, instead of having the main
thread stuck on a `wait-condition-variable'.
* module/ice-9/threads.scm (par-mapper): Add a `cons' argument; update
callers accordingly. Rewrite using nested futures.
* module/ice-9/futures.scm (%futures-waiting, %within-future?,
%future-prompt): New variables.
(let/ec): New macro.
(process-future!): Run FUTURE's thunk in a prompt; capture FUTURE's
continuation when it aborts, and add it to %FUTURES-WAITING. Set
%WITHIN-FUTURE? in the dynamic extent of the call FUTURE's thunk.
(process-futures): Move loop body to...
(process-one-future): ... here. New procedure.
(notify-completion): New procedure.
(touch)[work, loop]: New procedures.
When %WITHIN-FUTURE? and FUTURE is started, abort; if not
%WITHIN-FUTURE, call `work' while waiting.
When FUTURE is queued, call `work' too.
* test-suite/tests/future.test ("nested futures"): New tests.
* module/ice-9/futures.scm (<future>)[completion]: New field.
[done?]: Rename to...
[state]: ... this. Change `set-future-done?!' to
`set-future-state!', and `future-done?' to `future-state'.
(make-future): Initialize the `completion' field to 'queued.
(with-mutex): New macro.
(process-future!): Remove `set-future-done?!' call.
(process-futures): Check `future-state'. Unlock FUTURE's mutex before
processing it. Broadcast FUTURE's `completion' cond. var. when done.
(touch): Likewise.
* module/ice-9/futures.scm (%create-workers!): Use 'with-mutex' in case
an exception is thrown. Within the critical section, check to make
sure the worker pool hasn't already been created by another thread.
* module/ice-9/getopt-long.scm (%make-option-spec): Fix the formal
argument list to match the actual fields present.
(make-option-spec): Update to match the fixed %make-option-spec.
Reported by David Pirotte.
* module/ice-9/futures.scm (process-futures): Wait on %FUTURES-AVAILABLE
only when %FUTURES is empty.
The problem was obvious when running (begin (use-modules (ice-9
threads)) (par-map 1+ (iota 400000))) : eventually, only the main
thread would do the work, while the others would remain idle, waiting
on %FUTURES-AVAILABLE.
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
* libguile/private-options.h: Add SCM_CURLY_INFIX_P macro, and increment
SCM_N_READ_OPTIONS.
* libguile/read.c (sym_nfx, sym_bracket_list, sym_bracket_apply): New
variables.
(scm_read_opts): Add curly-infix reader option. Reformat to comply
with GNU coding standards.
(scm_t_read_opts): Add curly_infix_p and neoteric_p fields.
(init_read_options): Initialize new fields.
(CHAR_IS_DELIMITER): Add '{', '}', '[', and ']' as delimiters if
curly_infix_p is set.
(set_port_square_brackets_p, set_port_curly_infix_p): New functions.
(read_inner_expression): New function which contains the code that was
previously in 'scm_read_expression'. Handle curly braces when
curly_infix_p is set. If curly_infix_p is set and square_brackets_p
is unset, follow the Kawa convention: [...] => ($bracket-list$ ...)
(scm_read_expression): New function body to handle neoteric
expressions where appropriate.
(scm_read_shebang): Handle the new reader directives: '#!curly-infix'
and the non-standard '#!curly-infix-and-bracket-lists'.
(scm_read_sexp): Handle curly infix lists.
* module/ice-9/boot-9.scm (%cond-expand-features): Add srfi-105 feature
identifier.
* doc/ref/srfi-modules.texi (SRFI-105): Add stub doc for SRFI-105.
* doc/ref/api-evaluation.texi (Scheme Read): Add documentation for the
'curly-infix' read option, and the '#!curly-infix' and
'#!curly-infix-and-bracket-lists' reader directives.
* doc/ref/api-options.texi (Runtime Options): Add 'curly-infix' to the
list of read options.
* test-suite/Makefile.am: Add tests/srfi-105.test.
* test-suite/tests/srfi-105.test: New file.
* module/ice-9/boot-9.scm (define-option-interface): When printing
options help, e.g. for (read-options 'help), expand the width of the
first column by another tab stop, to accommodate option names of up to
23 characters.
* module/ice-9/regex.scm (fold-matches): Set regexp/notbol if the
starting position is nonzero.
* test-suite/tests/regexp.test (fold-matches): Check that when
matching /^foo/ against "foofoofoofoo", only one match results.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* module/ice-9/psyntax.scm (@): Return top-wrap instead of the wrap
applied to the '@' form, so that the symbol will be interpreted as a
top-level identifier and never refer to any lexical variable.
(@@): Change the syntax used to support R6RS 'library' forms to:
(@@ @@ (mod ...) body). Change the behavior of the documented
(@@ (mod ...) id) form to be the same as that of @, except for the use
of 'private' instead of 'public' in the psyntax mod: use syntax->datum
on the identifier, and return top-wrap instead of the wrap applied to
the '@@' form.
This fixes <http://bugs.gnu.org/10756> reported by Ludovic Courtès.
* module/ice-9/psyntax-pp.scm: Regenerate.
* module/ice-9/r6rs-libraries.scm (library): Use '@@ @@' syntax instead
of the older '@@' syntax.
* test-suite/tests/syncase.test (changes to expansion environment): Use
'@@ @@' syntax.
* module/Makefile.am: Add explicit dependencies for boot-9.go on the
files that it includes: quasisyntax.scm and r6rs-libraries.scm.
* module/ice-9/session.scm (procedure-arguments): Arrange to interpret
numbers in the "req" and "opt" positions of an 'arglist as N arguments
with unknown name.
* module/ice-9/eval.scm (primitive-eval): Set 'arglist on "complex"
procedures. Fixes http://bugs.gnu.org/10922.
* test-suite/tests/session.test ("procedure-arguments"): Add a test.
Fixes <http://bugs.gnu.org/10938>.
Based on a patch by Stefan Israelsson Tampe <stefan.itampe@gmail.com>.
* module/ice-9/session.scm (procedure-arguments): When the 'arglist
property is available, emit the `allow-other-keys?' pair. Use
`match-lambda'.
* test-suite/tests/session.test ("procedure-arguments")["aok? is
preserved"]: New test.
* module/ice-9/ftw.scm (scandir): Run the select? procedure on all
items, including subdirs and the `.' and `..' entries. Pass it the
basename of the file in question instead of the full name.
* test-suite/tests/ftw.test ("scandir"): Adapt expectation for the .test
selector. Add test for a selector that rejects everything.
* libguile/strports.c (scm_mkstrport): Remove initialization of
`pt->ilseq_handler'.
* module/ice-9/pretty-print.scm (truncated-print)[ellipsis]: Set
%DEFAULT-PORT-CONVERSION-STRATEGY to 'error.
* test-suite/tests/ports.test ("string
ports")["%default-port-conversion-strategy is honored"]: New test.
["wrong encoding"]: Rename to...
["wrong encoding, error"]: ... this. Explicitly set
%DEFAULT-PORT-CONVERSION-STRATEGY to 'error. Return #f when no
exception is raised.
This commit removes code that was newly deprecated in stable-2.0.
Conflicts:
libguile/deprecated.c
libguile/deprecated.h
libguile/modules.c
module/ice-9/boot-9.scm
module/ice-9/deprecated.scm
* libguile/frames.c (scm_frame_source, scm_frame_previous):
* libguile/stacks.c (scm_make_stack):
* module/ice-9/boot-9.scm (exception-printers):
* module/system/vm/frame.scm (frame-call-representation): Fix more
assumptions that frame-procedure is a program, or even a procedure.
* module/ice-9/vlist.scm (make-block): If we are making a hash table,
allocate it inline with the contents. Otherwise don't even add a
pointer to the block.
(block-hash-table?): New internal accessor.
(block-ref*): Remove. Vhash entries are no longer wrapped.
(block-ref):
(block-hash-table-next-offset):
(block-hash-table-set-next-offset!):
(block-hash-table-ref):
(block-hash-table-set!):
(block-hash-table-add!): Adapt to take content vector explicitly, and
to expect the hash table inline with the contents. Some of these
accessors are new. Adapt callers.
(assert-vlist): New helper.
(vlist-cons): Update comment.
(vhash?): Update scheme to allocate the hash table and chain links
inline with the contents.
(%vhash-fold*, %vhash-assoc): Rewrite to be more performant.
* module/ice-9/vlist.scm (set-block-next-free!): Define this instead of
increment-block-next-free!.
(block-append!): Refactor to take an offset, and only append if the
offset is the next free value, and there is space in the block.
(block-cons): Refactor to not be a loop. The partial evaluator would
have to understand effects analysis in order to be able to unroll it,
and there's at most one recursion.
Recovers the performance loss resulting from the previous commit.
* module/ice-9/vlist.scm: Use define-inlinable instead of define-inline,
to ensure strict argument evaluation. There is a slight performance
penalty, but I hope subsequent hacks make it up.
* module/ice-9/boot-9.scm (default-prompt-tag): Once parameters have
booted, redefine as a parameter.
(make-prompt-tag): Change from a gensym to a list. Thanks to Mark
Weaver for the suggestion.
* doc/ref/api-control.texi (Prompt Primitives): Update docs.
