Andy Wingo [Fri, 25 Mar 2011 14:35:20 +0000 (15:35 +0100)]
avoid running GC when SCM_I_CURRENT_THREAD is unset
* libguile/threads.c (guilify_self_1): Prevent finalizers from running
before SCM_I_CURRENT_THREAD is set.
(do_thread_exit_trampoline): Leave the thread in the registered state.
(on_thread_exit): Always unregister the thread here.
Andy Wingo [Fri, 25 Mar 2011 12:01:51 +0000 (13:01 +0100)]
threading / with_guile refactor to use more GC_stack_base
* libguile/init.h:
* libguile/init.c (scm_i_init_guile): Change arg to this internal
function from SCM_STACKITEM* to void*. Actually it's a
struct GC_stack_base*.
* libguile/bdw-gc.h: Don't do pthread redirects, because we don't want
to affect applications' pthread_* bindings.
* libguile/pthread-threads.h (scm_i_pthread_create)
(scm_i_pthread_detach, scm_i_pthread_exit, scm_i_pthread_cancel)
(scm_i_pthread_sigmask): Do pthread redirects here, in this internal
header.
* libguile/threads.h: Remove declaration of internal
scm_i_with_guile_and_parent. Remove declaration of undefined
scm_threads_init_first_thread. Make declaration of internal
scm_threads_prehistory actually internal, and take a void* (actually a
struct GC_stack_base*).
* libguile/threads.c (GC_get_stack_base): Implement a shim if this
function is unavailable, and fold in the implementations of
get_thread_stack_base.
(GC_call_with_stack_base): Actually implement.
(guilify_self_1): Take a GC_stack_base* as an arg.
(scm_i_init_thread_for_guile): Likewise, and set up libgc for
registration of other threads.
(scm_init_guile): Use GC_get_stack_base instead of our own guesswork.
(with_guile_and_parent, scm_i_with_guile_and_parent): Rework to
trampoline through a GC_call_with_stack_base.
(scm_threads_prehistory): Pass the "base" arg on to guilify_self_1.
Andy Wingo [Fri, 25 Mar 2011 09:47:10 +0000 (10:47 +0100)]
Revert "with-continuation-barrier carps, calls exit(3) _after_ unwinding"
This reverts commit
ecba00af6501e082b86c8f2f7730081c733509d7.
Andy Wingo [Thu, 24 Mar 2011 19:34:31 +0000 (20:34 +0100)]
bdw-gc 6.8 compatibility (hopefully)
* configure.ac (HAVE_GC_STACK_BASE): New check.
* libguile/threads.c (GC_UNIMPLEMENTED, GC_SUCCESS): Define if needed.
(GC_register_my_thread, GC_unregister_my_thread)
(GC_call_with_stack_base): Define shims if needed.
Andy Wingo [Thu, 24 Mar 2011 19:20:14 +0000 (20:20 +0100)]
with-continuation-barrier carps, calls exit(3) _after_ unwinding
* libguile/continuations.c (scm_handler, c_handler)
(scm_c_with_continuation_barrier, scm_with_continuation_barrier): Call
scm_handle_by_message_noexit in the post-unwind handler, so that
dynwinds
* test-suite/tests/continuations.test ("continuations"): Add a test.
Andy Wingo [Wed, 23 Mar 2011 16:05:28 +0000 (17:05 +0100)]
fix a failure to sync regs in vm bytevector ops
* libguile/vm-i-scheme.c (BV_SET_WITH_ENDIANNESS, BV_FIXABLE_INT_SET)
(BV_INT_SET, BV_FLOAT_SET): Sync registers before dispatching to the C
function.
Mark H Weaver [Tue, 22 Mar 2011 15:11:53 +0000 (11:11 -0400)]
Do not enter the debugger if the thrown key is in `pass-keys'
* module/system/repl/error-handling.scm (call-with-error-handling):
Do _not_ enter the debugger if the thrown key is in `pass-keys'.
Previously, for example, (throw 'quit) entered the debugger when run
from the REPL, despite the fact that 'quit is in `pass-keys'.
Ludovic Courtès [Sun, 20 Mar 2011 22:34:42 +0000 (23:34 +0100)]
Make VM string literals immutable.
* libguile/strings.c (scm_i_make_string, scm_i_make_wide_string): Add
`read_only_p' parameter. All callers updated.
* libguile/vm-i-loader.c (load_string, load_wide_string): Push read-only
strings.
* test-suite/tests/strings.test ("literals"): New test prefix.
BT Templeton [Sun, 20 Mar 2011 03:21:06 +0000 (23:21 -0400)]
fix guile-snarf
* libguile/snarf.h: New macro `SCM_SNARF_INIT_PREFIX'. (SCM_SNARF_INIT)
Use `SCM_SNARF_INIT_PREFIX' instead of including a literal marker. If
the preprocessor echoes #define directives to its output, this will
prevent `guile-snarf' from snarfing the `SCM_SNARF_INIT' definition
itself. Reported by Mike Gran <spk121@yahoo.com>.
* libguile/guile-snarf.in (modern_snarf): Don't output anything for
lines in which only one of the magic snarfing markers is present.
Modify the `sed' program for compatibility with POSIX `sed'. The new
`sed' program is based on a version by Wolfgang Jenkner
<wjenkner@inode.at>.
* test-suite/standalone/test-guile-snarf: New tests.
Andreas Rottmann [Sun, 20 Mar 2011 00:00:09 +0000 (01:00 +0100)]
Add VM test for call/cc in non-tail position
* test-suite/vm/t-call-cc.scm: Add test case using call/cc in a non-tail
position.
Andreas Rottmann [Sat, 19 Mar 2011 23:49:07 +0000 (00:49 +0100)]
Fix syntax error in benchmark-suite/Makefile.am
* benchmark-suite/Makefile.am (SCM_BENCHMARKS): Correct position of
a trailing backslash.
Michael Gran [Sat, 19 Mar 2011 22:25:28 +0000 (15:25 -0700)]
Benchmarks for string comparisons
* benchmark-suite/benchmarks/strings.bm: new file
* benchmark-suite/Makefile.am: add strings.bm
Andy Wingo [Sat, 19 Mar 2011 22:12:40 +0000 (23:12 +0100)]
fix new-frame push in call/cc
* libguile/vm-i-system.c (call/cc): Whoops, fix the new-frame push
here. A little birdie tells me a test case is coming soon.
Andy Wingo [Fri, 18 Mar 2011 12:18:47 +0000 (13:18 +0100)]
fix thread cleanup
* libguile/threads.h: Always declare a scm_i_thread_key, for cleanup
purposes, in the BUILDING_LIBGUILE case.
* libguile/threads.c (scm_i_thread_key): Init with a cleanup handler, so
any guile-specific info for a thread can be cleaned up reliably.
(guilify_self_1): Always set the thread key.
(do_thread_exit_trampoline, on_thread_exit): Enter guile-mode for the
guile-mode cleanup handler, and trampoline through a
gc_call_with_stack_base for reasons explained in the code.
(init_thread_key, scm_i_init_thread_for_guile): Always init the key.
(scm_i_with_guile_and_parent): No need for pthread_cancel cleanup
handlers, as the pthread key destructor will take care of that for
us.
(really_launch): Remove needless pthread_exit call with incorrect
comment.
Ludovic Courtès [Fri, 18 Mar 2011 10:24:51 +0000 (11:24 +0100)]
Fix `i18n.test' when the German or Greek locales aren't available.
* test-suite/tests/i18n.test (%german-utf8-locale, %greek-utf8-locale):
New variables.
(under-german-utf8-locale-or-unresolved,
under-greek-utf8-locale-or-unresolved): Use them.
Ludovic Courtès [Thu, 17 Mar 2011 21:44:25 +0000 (22:44 +0100)]
i18n: Add case mapping and case-insensitive string comparison tests.
Thanks to Mark H Weaver <mhw@netris.org> for coming up with most of the
examples.
* test-suite/tests/i18n.test (%german-utf8-locale-name,
%greek-utf8-locale-name): New variables.
(under-german-utf8-locale-or-unresolved,
under-greek-utf8-locale-or-unresolved): New procedures.
("text collation (German)", "text collation (Greek)"): New tests
prefixes.
("string mapping")["string-locale-upcase German",
"string-locale-upcase Greek", "string-locale-upcase Greek (two
sigmas)", "string-locale-downcase Greek", "string-locale-downcase
Greek (two sigmas)"]: New tests.
Ludovic Courtès [Thu, 17 Mar 2011 21:31:48 +0000 (22:31 +0100)]
i18n: Re-enable tests with the Turkish locale.
* test-suite/tests/i18n.test ("character mapping")["char-locale-upcase
Turkish", "char-locale-downcase Turkish"]: Re-enable. Passes with
GNU libc 2.12.1.
("string mapping")["string-locale-upcase Turkish",
"string-locale-downcase Turkish"]: Likewise.
Andy Wingo [Thu, 17 Mar 2011 17:53:11 +0000 (18:53 +0100)]
add multibyte regexp test
* test-suite/standalone/Makefile.am:
* test-suite/standalone/test-mb-regexp: New test, that the previous
patch fixed the abort() on fixup_multibyte_match.
Andy Wingo [Thu, 17 Mar 2011 17:29:08 +0000 (18:29 +0100)]
scm_{to,from}_locale_string use current locale, not current ports
* libguile/strings.c (scm_to_locale_stringn, scm_from_locale_stringn):
Use the encoding of the current locale, not of the current i/o ports.
Also use the current conversion strategy.
* doc/ref/api-data.texi (Conversion to/from C): Update docs.
Andy Wingo [Thu, 17 Mar 2011 11:33:58 +0000 (12:33 +0100)]
allow ,option on-error report instead of debug
* module/system/repl/command.scm:
* module/system/repl/debug.scm (terminal-width): Move terminal-width
here, make it thread-local, and export it.
(print-locals, print-frame, print-frames): Default width to
terminal-width.
* module/system/repl/error-handling.scm (call-with-error-handling): Add
`report' and `backtrace' on-error handlers.
* module/system/repl/common.scm (repl-default-options): Add on-error
REPL option, defaulting to `debug', but which may be changed.
* module/system/repl/repl.scm (run-repl): Pass the #:on-error REPL
option to call-with-error-handling.
Andy Wingo [Thu, 17 Mar 2011 10:43:06 +0000 (11:43 +0100)]
add heap-allocated-since-gc to gc-stats
* libguile/gc.c (scm_gc_stats): Use add bytes_since_gc to the alist,
under "heap-allocated-since-gc", and remove dead code.
Andy Wingo [Thu, 17 Mar 2011 10:42:50 +0000 (11:42 +0100)]
fix code that causes warnings on gcc 4.6
* libguile/arrays.c (scm_i_read_array):
* libguile/backtrace.c (display_backtrace_body):
* libguile/filesys.c (scm_readdir)
* libguile/i18n.c (chr_to_case):
* libguile/ports.c (register_finalizer_for_port):
* libguile/posix.c (scm_nice):
* libguile/stacks.c (scm_make_stack): Clean up a number of
set-but-unused vars. Thanks to Douglas Mencken for the report.
* libguile/numbers.c (scm_log, scm_exp): Fix a few #if cases that should
be #ifdef.
Andy Wingo [Thu, 17 Mar 2011 09:39:02 +0000 (10:39 +0100)]
add pointer->scm, scm->pointer
* libguile/foreign.c (scm_pointer_to_scm, scm_scm_to_pointer): New
functions, useful to pass and receive SCM values to and from foreign
functions.
* module/system/foreign.scm: Export the new functions.
* doc/ref/api-foreign.texi (Foreign Variables): Add docs.
* test-suite/tests/foreign.test ("pointer<->scm"): Tests.
Andy Wingo [Tue, 15 Mar 2011 22:54:06 +0000 (23:54 +0100)]
add more prompt/abort tests
* test-suite/tests/control.test: Use c&e tests for most test blocks.
Note that this did not catch the recent bug.
("reified continuations"): Add a new test for capturing partial
continuations containing pending call frames. Before these would
contain dynamic links pointing out of the continuation segment, which
would not be relocated; now, the dynamic links are only made when the
frames are activated.
Thanks to Wolfgang J Moeller for the bug report and test case.
Andy Wingo [Tue, 15 Mar 2011 22:33:32 +0000 (23:33 +0100)]
fix frame dynamic linkage in the face of partial continuation application
* libguile/vm-i-system.c (new-frame): Though it was appealing to set the
dynamic link here on the incomplete frame, we no longer do that, for
the reasons mentioned in the code.
(call, mv-call): Adapt to set the frame's dynamic link.
* libguile/vm-engine.c (vm_engine): Don't set dynamic link here, even
for boot program.
* libguile/frames.c (scm_frame_num_locals, scm_frame_local_ref)
(scm_frame_local_set_x): Fix up not-yet-active frame detection.
Ludovic Courtès [Sun, 13 Mar 2011 22:21:07 +0000 (23:21 +0100)]
doc: Remove "lack of support for Unicode I/O and strings".
* doc/ref/api-io.texi (R6RS I/O Ports): Remove 1.8ish comment.
Andreas Rottmann [Sun, 13 Mar 2011 22:14:10 +0000 (23:14 +0100)]
Enhance transcoder-related functionality of `(rnrs io ports)'
* module/rnrs/io/ports.scm (transcoder-eol-style)
(transcoder-error-handling-mode): Export these.
(textual-port?): Implement this procedure and export it.
* module/rnrs.scm: Export these here as well.
* module/rnrs/io/ports.scm (port-transcoder): Implement this procedure.
(binary-port?): Treat only ports without an encoding as binary ports,
add docstring.
(standard-input-port, standard-output-port, standard-error-port):
Ensure these are created without an encoding.
(eol-style): Add `none' as enumeration member.
(native-eol-style): Switch to `none' from `lf'.
* test-suite/tests/r6rs-ports.test (7.2.7 Input ports)
(8.2.10 Output ports): Test binary-ness of `standard-input-port',
`standard-output-port' and `standard-error-port'.
(8.2.6 Input and output ports): Add test for `port-transcoder'.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Andreas Rottmann [Sun, 13 Mar 2011 21:39:26 +0000 (22:39 +0100)]
Export `current-*-port' from `(rnrs io ports)'
* module/rnrs/io/ports.scm: Export `current-input-port',
`current-output-port' and `current-error-port' (see R6RS 8.2.7 "Input
ports" and 8.2.10 "Output ports").
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Andreas Rottmann [Sun, 13 Mar 2011 21:39:14 +0000 (22:39 +0100)]
Add `get-string-n' and `get-string-n!' for R6RS ports
* libguile/r6rs-ports.c (scm_get_string_n_x): Implement `get-string-n!'
in C for efficiency.
* libguile/r6rs-ports.h: Add prototype for this function.
* module/ice-9/binary-ports.scm: Export `get-string-n!'.
* module/rnrs/io/ports.scm (get-string-n): Implement based on
`get-string-n!'.
Export both `get-string-n!' and `get-string-n'.
* module/rnrs.scm: Also export these.
* test-suite/tests/r6rs-ports.test (8.2.9 Textual input): Add a few
tests for `get-string-n' and `get-string-n!'.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Ludovic Courtès [Sun, 13 Mar 2011 15:09:55 +0000 (16:09 +0100)]
Work around weak-value hash table bug in `define-wrapped-pointer-type'.
* module/system/foreign.scm (define-wrapped-pointer-type)[wrap]: Use
`hash-ref' and `hash-set!' instead of `hash-create-handle!' and
`set-cdr!'.
Ludovic Courtès [Fri, 11 Mar 2011 20:01:35 +0000 (21:01 +0100)]
Fix `define-inlinable' in SRFI-9 so that arguments are evaluated only once.
* module/srfi/srfi-9.scm (define-inlinable): When inlining, evaluate the
arguments only once. Reported by Andreas Rottmann; thanks to Andy
Wingo for the elegant solution.
* test-suite/tests/srfi-9.test ("side-effecting arguments"): New test
prefix.
Andy Wingo [Fri, 11 Mar 2011 19:53:13 +0000 (20:53 +0100)]
fix port-filename without readline to match the docs
* libguile/init.c (stream_body, scm_standard_stream_to_port): Don't name
stdin, stdout, and stderr -- at least not as strings. That confuses
any code which tries to treat port-filename as a real filename, like
the syntax expander, or the `load' procedure/macro. Also this
behavior matches the docs now.
Mark H Weaver [Thu, 10 Mar 2011 22:35:19 +0000 (17:35 -0500)]
Fix bug to make `string=' much faster
* libguile/srfi-13.c (scm_string_eq): Fix a bug which caused the slow
general string_compare function to be used for strings of unequal
lengths.
Ludovic Courtès [Thu, 10 Mar 2011 21:24:40 +0000 (22:24 +0100)]
Thanks, Aidan.
Ludovic Courtès [Thu, 10 Mar 2011 21:24:23 +0000 (22:24 +0100)]
FFI: Return the right alignment for structures.
* libguile/foreign.c (scm_alignof): Fix handling of structure alignment.
Reported by Aidan Gauland <aidalgol@no8wireless.co.nz>.
* test-suite/tests/foreign.test ("structs")["alignof { int8, double,
int8 }", "int8, { int8, double, int8 }, int16"]: New tests.
Mark H Weaver [Mon, 7 Mar 2011 11:27:42 +0000 (06:27 -0500)]
Update Gnulib; add new modules; remove `round' module.
This updates Gnulib to
v0.0-4951-g6ff7b70.
* m4/gnulib-cache.m4: Add floor, ceil, frexp, and ldexp. Add wchar as
an explicit dependency; it had been present as an indirect dependency
before, but no longer. Remove round, which I had requested earlier,
but turned out to be unnecessary.
Andy Wingo [Wed, 9 Mar 2011 21:37:53 +0000 (22:37 +0100)]
fix-letrec tweaks
* module/language/tree-il/fix-letrec.scm (partition-vars): Previously,
for letrec* we treated all unreferenced vars as complex, because of
orderings of effects that could arise in their definitions. But we
can actually keep simple and lambda vars as unreferenced, as their
initializers cannot cause side effects.
(fix-letrec!): Remove letrec* -> letrec code, as it's unneeded.
Andreas Rottmann [Wed, 9 Mar 2011 20:36:54 +0000 (21:36 +0100)]
Don't mix definitions and expressions in SRFI-9
The expansion of `define-inlinable' contained an expression, which made
SRFI-9's `define-record-type' fail in non-toplevel contexts ("definition
used in expression context").
* module/srfi/srfi-9.scm (define-inlinable): Get rid of apparently
useless expression in the expansion, so the expansion yields only
definitions. At the same time, use a space in the generated names to
lessen the chances of name conflicts, also avoiding -Wunused-toplevel
warnings.
* test-suite/tests/srfi-9.test (non-toplevel): New test verifying that
`define-record-type' works in non-toplevel context as well.
* doc/ref/srfi-modules.texi (SRFI-9 - define-record-type): Add
subsubsection noting that Guile does not enforce top-levelness.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Mark H Weaver [Wed, 9 Mar 2011 06:14:43 +0000 (01:14 -0500)]
Add scm_from_latin1_keyword and scm_from_utf8_keyword
* libguile/keywords.c (scm_from_latin1_keyword, scm_from_utf8_keyword):
New functions appropriate for use when keyword name is a constant.
(scm_from_locale_keyword, scm_from_locale_keywordn): Change formal
parameter from `str' to `name'.
* libguile/keywords.h: Add prototypes for new functions. Change formal
parameter of scm_from_locale_keyword* from `str' to `name'.
* doc/ref/api-data.texi: Document new functions. Remind users that
scm_from_locale_keyword should not be used when the name is a C string
constant. Change formal parameter from `str' to `name'.
Mark H Weaver [Tue, 8 Mar 2011 18:13:54 +0000 (13:13 -0500)]
Improve docs of string and symbol conversions from C strings
* doc/ref/api-data.texi (Conversion to/from C): Document
scm_from_latin1_string, scm_from_utf8_string, and
scm_from_utf32_string. Remind readers that these functions should be
used to convert C string constants, and that scm_from_locale_string is
_not_ appropriate for that purpose.
(Symbol Primitives): Document scm_from_latin1_symbol and
scm_from_utf8_symbol. Remind readers that these functions should be
used when the specified names are C string constants, and that
scm_from_locale_symbol is _not_ appropriate for that purpose.
Mark H Weaver [Fri, 4 Mar 2011 18:44:02 +0000 (13:44 -0500)]
Within `while', `continue' takes zero arguments
* module/ice-9/boot-9.scm (while): Report an error if `continue' is
passed one or more arguments. Previously, it would report an error if
`(continue arg rest ...)' was found within the `while', but not if
`continue' was found bare and later applied to one or more arguments,
e.g. `(apply continue (list arg rest ...))'.
Mark H Weaver [Mon, 7 Mar 2011 01:27:40 +0000 (20:27 -0500)]
Quotient, remainder and modulo accept inexact integers
* libguile/numbers.c (scm_quotient, scm_remainder, scm_modulo): Accept
inexact integers as well as exact ones, as required by the R5RS.
* test-suite/tests/numbers.test (quotient, remainder, modulo): Add tests.
Mark H Weaver [Tue, 1 Mar 2011 18:46:08 +0000 (13:46 -0500)]
Remove incorrect footnote from GOOPS manual
* doc/ref/goops.texi (Inheritance): Remove footnote which incorrectly
stated that <complex> was not shown in the class hierarchy figure.
Mark H Weaver [Wed, 23 Feb 2011 02:39:37 +0000 (21:39 -0500)]
Update comments regarding GMP earlier than 4.2.
* libguile/numbers.c: Update comments regarding GMP earlier than 4.2.
Remove speculations about versions of GMP that had not yet been
released when the comments were written. Replace them with facts that
are now known about the changes made in GMP 4.2.
Mark H Weaver [Tue, 1 Mar 2011 17:37:01 +0000 (12:37 -0500)]
Fix description of the R6RS `finite?' in manual
* doc/ref/r6rs.texi (rnrs base): `(finite? x)' returns true iff x is
neither infinite nor a NaN. Previously, it stated that `finite?' was
the negation of `infinite?', which was incorrect because NaNs are
neither finite nor infinite. Combine description of 'nan?' with those
of `finite?' and `infinite?'.
Mark H Weaver [Tue, 1 Mar 2011 17:46:38 +0000 (12:46 -0500)]
Fix bytevectors VALIDATE_REAL to test for reals, not rationals
Reported and fixed by Daniel Llorens <dll@bluewin.ch>.
* libguile/bytevectors.c (VALIDATE_REAL): Test for reals, not rationals.
* test-suite/tests/srfi-4.test (f32 vectors, f64 vectors): Add tests.
Andy Wingo [Tue, 8 Mar 2011 21:34:53 +0000 (22:34 +0100)]
scm_public_ref et al docs
* doc/ref/api-modules.texi (Accessing Modules from C): Add docs for the
new C procedures.
Andy Wingo [Tue, 8 Mar 2011 20:53:02 +0000 (21:53 +0100)]
eval-string docs
* doc/ref/api-evaluation.texi (Fly Evaluation): Update eval-string
documentation.
Andy Wingo [Tue, 8 Mar 2011 20:06:12 +0000 (21:06 +0100)]
document scm_call_{5,6,n}
* doc/ref/api-evaluation.texi (Fly Evaluation): Document
scm_call_{5,6,n}.
Andy Wingo [Tue, 8 Mar 2011 19:57:41 +0000 (20:57 +0100)]
add scm_call_{5,6}
* libguile/eval.h:
* libguile/eval.c (scm_call_5, scm_call_6): New scm_call functions; why
not.
Andy Wingo [Tue, 8 Mar 2011 08:30:33 +0000 (09:30 +0100)]
fix scm_setter
* libguile/procs.c (scm_setter): Only get at the setter slot if the pure
generic actually has a setter. Needs test.
* test-suite/tests/goops.test ("defining generics"):
("defining accessors"): Add `setter' tests.
Andy Wingo [Tue, 8 Mar 2011 08:29:24 +0000 (09:29 +0100)]
core eval-string uses (ice-9 eval-string)
* libguile/strports.c (scm_eval_string_in_module): Use eval-string from
(ice-9 eval-string).
Andy Wingo [Tue, 8 Mar 2011 08:27:23 +0000 (09:27 +0100)]
add scm_c_public_ref et al
* libguile/modules.h:
* libguile/modules.c (scm_public_lookup, scm_private_lookup)
(scm_c_public_lookup, scm_c_private_lookup, scm_public_ref)
(scm_private_ref, scm_c_public_ref, scm_c_private_ref)
(scm_public_variable, scm_private_variable, scm_c_public_variable)
(scm_c_private_variable): New helpers to get at variables and values
in modules.
Ludovic Courtès [Sun, 6 Mar 2011 22:02:57 +0000 (23:02 +0100)]
Handle `letrec*' like `letrec' in simple cases.
* module/language/tree-il/fix-letrec.scm (fix-letrec!): When X is a
`letrec*' with only lambdas and simple expressions, analyze it as if
it were a `letrec'.
* test-suite/tests/tree-il.test ("letrec"): Add test for
`(letrec* (x y) (xx yy) ((const 1) (const 2)) (lexical y yy))'.
Ludovic Courtès [Sun, 6 Mar 2011 21:27:53 +0000 (22:27 +0100)]
Have `gc-profile.scm' make sure it's on a Linux-based system.
* gc-benchmarks/gc-profile.scm (memory-mappings): Check %HOST-TYPE for
"-linux-".
Ludovic Courtès [Sun, 6 Mar 2011 21:26:49 +0000 (22:26 +0100)]
Add `gc-benchmarks/' to the distribution.
* gc-benchmarks/Makefile.am: New file.
* configure.ac: Produce it.
* Makefile.am (SUBDIRS): Add `gc-benchmarks'.
Ludovic Courtès [Sun, 6 Mar 2011 21:13:10 +0000 (22:13 +0100)]
Make `object->string' explicitly close its string output port.
* libguile/strports.c (scm_object_to_string): Close PORT before
returning the resulting string.
Ludovic Courtès [Sun, 6 Mar 2011 20:47:48 +0000 (21:47 +0100)]
Slightly optimize `gensym'.
* libguile/symbols.c (default_gensym_prefix): New variable.
(scm_gensym): Use it. Use `scm_from_latin1_stringn' instead of
`scm_from_locale_stringn'.
(scm_init_symbols): Initialize DEFAULT_GENSYM_PREFIX.
Ludovic Courtès [Sun, 6 Mar 2011 13:31:28 +0000 (14:31 +0100)]
Simply grow string port buffers geometrically.
* libguile/strports.c (SCM_WRITE_BLOCK): Remove.
(st_flush): Multiply `pt->write_buf_size' by 2.
(st_seek): Likewise when TARGET == PT->write_buf_size.
Ludovic Courtès [Sun, 6 Mar 2011 10:42:37 +0000 (11:42 +0100)]
Let `scm_mkstrport' allocate buffers on the caller's behalf.
* libguile/strports.c (INITIAL_BUFFER_SIZE): New macro.
(scm_mkstrport): If STR is false, allocate a bytevector on the
caller's behalf.
(scm_object_to_string, scm_call_with_output_string,
scm_open_output_string): Pass SCM_BOOL_F as the STR argument of
`scm_mkstrport'.
* libguile/backtrace.c (scm_display_application,
display_backtrace_body): Likewise.
* libguile/gdbint.c (scm_init_gdbint): Likewise.
* libguile/print.c (scm_simple_format): Likewise.
Ludovic Courtès [Sat, 5 Mar 2011 19:15:09 +0000 (20:15 +0100)]
Use a bytevector as the backing buffer of string ports.
* libguile/strports.c (st_resize_port): Adjust to deal with OLD_STREAM
and NEW_STREAM as bytevectors.
(scm_mkstrport): Store a bytevector in the port's stream rather than a
string.
Andy Wingo [Sat, 5 Mar 2011 22:16:11 +0000 (23:16 +0100)]
add ice-9 eval-string
* module/Makefile.am:
* module/ice-9/eval-string.scm: New module, for use in implementing the
scm_c_eval_string_from_file_line suggestion.
* test-suite/Makefile.am:
* test-suite/tests/eval-string.test: New tests.
Andy Wingo [Sat, 5 Mar 2011 20:48:47 +0000 (21:48 +0100)]
remove obsolete comments
* libguile/eval.c (scm_nconc2last):
* libguile/strports.c (scm_c_read_string): Remove some obsolete
comments.
Mark Harig [Sat, 5 Mar 2011 00:36:28 +0000 (19:36 -0500)]
Updated Guile manual page.
* doc/guile.1: Added the current month and year, Guile version
descriptive text, and the text GNU to the title.
Updated the nroff formatting commands for the SYNOPSIS and OPTIONS
sections to what 'man' prescribes. See 'man(1)', 'man(7)', and
'man-pages(7)'.
Corrected grammar, spelling, and capitalization (for example,
'scheme' to 'Scheme').
Vertical white-space was non-standard (two lines between some
sections, one space between others). Changed this to the standard
one empty line before each section heading, and added dots (a single
period on a line) before every section heading (.SH) so that
maintainers will find the readability unchanged.
Added white space to follow the 'groff' recommendation of starting
every sentence on its own line, and breaking sentences at
punctuation.
Corrected an error in description of the info command.
Added the missing option '--no-debug', and the short switches '-h'
and '-v'.
Changed the description of the environment variable
GUILE_LOAD_COMPILED_PATH so that it references the Guile variable
`%load-compiled-path' instead of the variable `%load-path'.
Updated the copyright to include 2011.
BT Templeton [Tue, 22 Feb 2011 18:15:31 +0000 (13:15 -0500)]
guile-snarf: allow multiple init actions on one line
* libguile/guile-snarf.in (modern_snarf): Allow programs to specify
multiple initialization actions on a single line. This makes it
possible for C programs to define multiple subrs with a single macro
invocation.
* test-suite/standalone/test-guile-snarf: Enable more tests.
BT Templeton [Tue, 22 Feb 2011 20:15:33 +0000 (15:15 -0500)]
add guile-snarf tests
* test-suite/standalone/test-guile-snarf: New file.
* test-suite/standalone/Makefile.am: Add `test-guile-snarf'.
Andy Wingo [Fri, 4 Mar 2011 10:16:15 +0000 (11:16 +0100)]
repl: terminal-width by default
* module/system/repl/command.scm (terminal-width): New parameter that
will use the true terminal width if unset.
(backtrace, locals): Default to (terminal-width).
(width): Simplify.
Michael Gran [Mon, 21 Feb 2011 05:53:46 +0000 (21:53 -0800)]
Add ,width meta-command to set screen width in debug output
This meta-command allows one to set the default number of columns
that output from ,backtrace and ,locals shall occupy.
* doc/ref/scheme-using.texi (Debug Commands): document ,width
* module/system/repl/command.scm (*width*): new var
(backtrace, locals): use *width* in optarg
(width): new meta-command
Andy Wingo [Fri, 4 Mar 2011 09:33:51 +0000 (10:33 +0100)]
fix ,stat
* module/system/repl/command.scm (statistics): Fix for BDW-GC.
Unfortunately we still don't have mallocation or time taken.
Andy Wingo [Thu, 3 Mar 2011 22:51:20 +0000 (23:51 +0100)]
repl.scm understands comments
* module/system/repl/repl.scm (read-comment, read-scheme-line-comment)
(read-scheme-datum-comment): New helpers.
(meta-reader): Take a language instead of a reader. If we have a
nonwhitespace char, first check to see that it's a comment, and if so,
read it off and loop.
(prompting-meta-read): Call meta-reader with the lang.
Andy Wingo [Thu, 3 Mar 2011 22:19:35 +0000 (23:19 +0100)]
repl.scm refactor
* module/system/repl/repl.scm (flush-leading-whitespace): Rename from
next-char.
(meta-reader): Use flush-leading-whitespace.
(run-repl): Use flush-to-newline after the evaluation, which seems to
be the same as what we did before.
Andy Wingo [Thu, 3 Mar 2011 11:46:49 +0000 (12:46 +0100)]
fix encoding scanning for non-seekable ports
* libguile/read.c (scm_i_scan_for_encoding): If possible, just use the
read buffer for the encoding scan, and avoid seeking. Fixes
`(open-input-file "/dev/urandom")', because /dev/urandom can't be
seeked backwards.
Andy Wingo [Thu, 3 Mar 2011 10:29:27 +0000 (11:29 +0100)]
more module-use-interfaces! tweaks
* module/ice-9/boot-9.scm (module-use-interfaces!): Fix up to prevent
duplication in the use list of multiple incoming interfaces.
* test-suite/tests/modules.test ("module-use"): Add tests.
Andreas Rottmann [Thu, 3 Mar 2011 10:09:54 +0000 (11:09 +0100)]
Use module identity to filter for existing modules
This fixes a problem with R6RS's `import' in particuliar: when importing
a subset of a library/module, the interface created for that purpose
inherits the name of the module it is derived from. The low-level
primitives that are used for importing would then disregard earlier
imports from the same module.
An example for this bug can be seen with the following library
definition:
(library (test-guile2)
(export foo)
(import (only (rnrs base) define)
(only (rnrs base) error))
(define (foo . args)
#t))
In the above, the import of `define' would be disregarded when `error'
is imported, thus leading to a syntax error, since `(foo . args)' is
treated as an application, since the binding of `define' would be not
present.
* module/ice-9/boot-9.scm (module-use!): Remove the filtering of the
existing imports of the module by name; a check for identity is
already done beforehand.
(module-use-interfaces!): Filter the existing imports by identity
instead of filtering them by their names.
Ludovic Courtès [Wed, 2 Mar 2011 23:13:30 +0000 (00:13 +0100)]
Remove extra comma after `SCM_ARRAY_ELEMENT_TYPE_LAST'.
* libguile/array-handle.h (scm_t_array_element_type): Remove extra comma
after last element. Reported by David Fang <fang@csl.cornell.edu>.
Indent.
Ludovic Courtès [Wed, 2 Mar 2011 21:12:56 +0000 (22:12 +0100)]
Improve the documentation for `dynamic-link'.
* doc/ref/api-foreign.texi (Foreign Libraries): Make it clear that the
LIBRARY argument of `dynamic-link' should not contain an extension.
(Foreign Functions): Add cross-reference from `load-extension' to
`dynamic-link'. Typeset file names and module names correctly.
Mark H Weaver [Wed, 2 Mar 2011 11:02:58 +0000 (06:02 -0500)]
Portability fixes for popen.test (for when /bin/sh is not bash)
* test-suite/tests/popen.test (open-input-pipe no-duplicate): Pass
"read REPLY" command instead of "read" to the subshell, for improved
portability. In particular, it is needed when /bin/sh is dash.
(open-output-pipe no-duplicate): Pass "exec guile [...]" instead of
"guile [...]" to the subshell, to ensure that the subshell will not
run guile as a subprocess while holding a duplicate of STDIN, which
would cause this test to fail. This is needed when /bin/sh is dash.
Ludovic Courtès [Mon, 28 Feb 2011 22:33:47 +0000 (23:33 +0100)]
Have `read' update line/column numbers when reading SCSH block comments.
* libguile/read.c (scm_read_scsh_block_comment): Use `scm_getc' instead
of `scm_get_byte_or_eof'.
* test-suite/tests/reader.test ("read-options")["position of SCSH block
comment"]: New test.
Andy Wingo [Mon, 28 Feb 2011 19:53:40 +0000 (20:53 +0100)]
update port-filename docs
* doc/ref/api-io.texi (File Ports):
* libguile/ports.c (scm_port_filename): Fix docs to match
implementation.
Ludovic Courtès [Sun, 27 Feb 2011 23:40:45 +0000 (00:40 +0100)]
Link stand-alone tests against libgc.
Reported by Andreas Rottmann <a.rottmann@gmx.at>.
* test-suite/standalone/Makefile.am (LIBGUILE_LDADD): New variable.
(test_num2integral_LDADD, test_round_LDADD, libtest_asmobs_la_LIBADD,
libtest_ffi_la_LIBADD, test_list_LDADD, test_unwind_LDADD,
test_conversion_LDADD, test_loose_ends_LDADD, test_scm_c_read_LDADD,
test_scm_take_locale_symbol_LDADD, test_scm_take_u8vector_LDADD,
libtest_extensions_la_LIBADD, test_with_guile_module_LDADD,
test_scm_with_guile_LDADD): Use it.
Ludovic Courtès [Sun, 27 Feb 2011 23:21:48 +0000 (00:21 +0100)]
Strip any CPPFLAGS other than `-I' from `guile-2.0.pc'.
* configure.ac: Strip anything beyond `-I' from $GUILE_CFLAGS so that
`guile-2.0.pc' does not export them to the user. Reported and fixed
by Bruno Haible <bruno@clisp.org>.
Andy Wingo [Sun, 27 Feb 2011 22:26:08 +0000 (23:26 +0100)]
flush all input on a read error
* module/system/repl/repl.scm (flush-all-input): New helper.
(prompting-meta-read): Flush all input on a read error, as we could be
within some expression or a string or something.
Andy Wingo [Sun, 27 Feb 2011 22:15:13 +0000 (23:15 +0100)]
scm-error-printer resilience
* module/ice-9/boot-9.scm (scm-error-printer): Allow #f for rest args,
interpreting it as '(). Fixes regexp throws, which are of the form:
(regular-expression-syntax "make-regexp" "Invalid preceding regular expression" #f ("?.*"))
Andy Wingo [Sun, 27 Feb 2011 12:07:04 +0000 (13:07 +0100)]
update R6RS incompatibilities
* doc/ref/r6rs.texi (R6RS Incompatibilities): Update.
Andy Wingo [Sun, 27 Feb 2011 11:59:44 +0000 (12:59 +0100)]
regenerate psyntax-pp
* module/ice-9/psyntax-pp.scm: Regenerate.
Andy Wingo [Sun, 27 Feb 2011 11:58:54 +0000 (12:58 +0100)]
add syncase test
* test-suite/tests/syncase.test ("top-level expansions"): New test.
Andy Wingo [Sun, 27 Feb 2011 11:48:23 +0000 (12:48 +0100)]
chi-top-sequence defines macros before expanding other exps
* module/ice-9/psyntax.scm (chi-top-sequence): Manually inline
eval-if-c&e into its two call sites; I found it hard to understand
otherwise. If the mode is just 'e, defer expansion of definitions and
expressions until the end, so that they can be expanded in a context
of all syntax expanders defined in the sequence.
Andy Wingo [Sun, 27 Feb 2011 11:07:48 +0000 (12:07 +0100)]
psyntax: fold chi-top-sequence into chi-top
* module/ice-9/psyntax.scm (chi-top-sequence): Pull chi-top into the
body of this toplevel begin expander. This will let us do r6rs
toplevel expansion correctly.
(chi-top): Remove.
(macroexpand): Dispatch to chi-top-sequence directly.
Ludovic Courtès [Fri, 25 Feb 2011 13:54:36 +0000 (14:54 +0100)]
Fix `gc-profile.scm'.
* gc-benchmarks/gc-profile.scm (memory-mappings)[mapping-line-rx]: Fix
and give an example.
(total-heap-size): Fix docstring.
Andy Wingo [Fri, 25 Feb 2011 09:48:35 +0000 (10:48 +0100)]
make-weak-key-hash-table vacuuming
* libguile/hashtab.c (scm_make_weak_key_hash_table): Whoops, fix the
case I actually cared about.
Ludovic Courtès [Thu, 24 Feb 2011 22:17:23 +0000 (23:17 +0100)]
Revert ""latin1" -> "Latin-1"."
This reverts commit
c2c550ca9d2442d070f79ed8bacb8db173c72df3.
The name "latin1" is standardized by IANA, unlike the other one.
Reported by Bruno Haible.
Ludovic Courtès [Thu, 24 Feb 2011 22:17:06 +0000 (23:17 +0100)]
Make `locale-digit-grouping' more robust.
* libguile/i18n.c (scm_nl_langinfo)[GROUPING]: Consider negative numbers
like `CHAR_MAX'. Reported by David Fang <fang@csl.cornell.edu>.
Fix suggested by Bruno Haible <bruno@clisp.org>.
Ludovic Courtès [Thu, 24 Feb 2011 22:13:54 +0000 (23:13 +0100)]
Fix README.
* README: Remove mention of an alpha release. Reported by
Mark H. Weaver.
Andy Wingo [Thu, 24 Feb 2011 16:00:30 +0000 (17:00 +0100)]
weak hash tables vacuum stale entries after a gc
* libguile/hashtab.c (scm_c_register_weak_gc_callback): New private
helper, arranges for a C function to be called with a SCM as an
argument, as long as the argument is reachable by GC.
(scm_make_weak_key_hash_table)
(scm_make_weak_value_hash_table)
(scm_make_doubly_weak_hash_table): Register a weak GC callback to
vacuum_weak_hash_table.
Andy Wingo [Thu, 24 Feb 2011 15:30:08 +0000 (16:30 +0100)]
re-enable the after-gc-hook
* libguile/gc.c (scm_gc): No need to take a mutex here. Don't run the
hook, the hook will run itself.
(scm_c_register_gc_callback): New private helper, registers a callback
the next time GC happens.
(system_gc_callback): Guile's internal callback that runs
scm_after_gc_c_hook, which itself queues a call to the after-gc-hook.
(scm_storage_prehistory): Queue up a call to system_gc_callback.
Andy Wingo [Thu, 24 Feb 2011 12:12:58 +0000 (13:12 +0100)]
pointerless backing buffers for string ports
* libguile/strports.c (scm_mkstrport): String port string buffer
allocated atomically.
Andy Wingo [Thu, 24 Feb 2011 12:10:16 +0000 (13:10 +0100)]
errno saving in display_string
* libguile/print.c (display_string): Fix a case in which perhaps `errno'
could have been stompled.
Andy Wingo [Thu, 24 Feb 2011 10:10:19 +0000 (11:10 +0100)]
web server more assiduous about closing ports
* module/web/uri.scm:
* module/web/server.scm (call-with-output-string*):
(call-with-output-bytevector*): Local procs to output to strings or
bytevectors, *and then close the port*. We can't make this change in
call-with-output-string because it would be incompatible.
* module/web/uri.scm (call-with-encoded-output-string, decode-string)
(uri-decode)
* module/web/server.scm (call-with-encoded-output-string): Use the new
helpers.
Andy Wingo [Wed, 23 Feb 2011 20:04:26 +0000 (21:04 +0100)]
open-pipe* pumps pipes guardian
* module/ice-9/popen.scm (open-pipe*): Hack around the lack of an
after-gc hook, and pump the pipes guardian here in the procedure that
adds to the guardian.
Andy Wingo [Wed, 23 Feb 2011 10:59:38 +0000 (11:59 +0100)]
GC dead links in weak hash tables before a possible rehash
* libguile/hashtab.c (vacuum_weak_hash_table): New helper, goes through
the entirety of a weak hash table, vacuuming dead entries.
(scm_hash_fn_create_handle_x): If when adding to a weak hash table, we
would trigger a rehash, vacuum the table first. The weak_bucket_assoc
would have only caught dead entries within one bucket.
Without this patch, the following code leaks:
(let lp ()
(call-with-output-string
(lambda (port)
(display "foo" port)))
(lp))