The new non-libltdl foreign library loading algorithm from 3.0.6
fails to cover common cases regarding how libtool names and installs
DLL files. Notably, it fails to recognize when libtool has added the
major version number into the filename itself, such as libfoo-1.dll
Also, it does not search in binary directories and the PATH for DLL
files, where libtool is likely to install DLLs.
This adds the option to search for dlls with major version numbers
in the filename, and modifies the search strategy for DLL-using
OSs to check bindir and PATH.
For MSYS, libraries are installed with the 'msys-' prefix. So this
modifies load-foreign-library to handle that prefix as well.
It changes the #:rename-on-cygwin? option to #:host-type-rename? to
better reflect that is works on both Cygwin and MSYS.
Partially based on a patch by Hannes Müller.
* NEWS: updated
* doc/ref/api-foreign.texi: document updates to load-foreign-library
and system-dll-path
* module/system/foreign-library.scm (is-integer-string?): new utility function
(dll-name-match?): new utility function
(find-best-dll-from-matches): new utility function
(dll-exists-with-version): new function that implements new dll search logic
(file-exists-with-extension): add flag argument to allow new dll search
(file-exists-in-path-with-extension): add flag argument to all new dll search
(system-dll-path): new parameter
(lib->msys): new helper function
(load-foreign-library): add new optarg flag #:allow-dll-version-suffix?
Pass new flag to library search functions.
Implement new search strategy for #:search-system-paths? on DLL systems'
replace #:rename-on-cygwin? with #:host-type-rename?
Use that option to rename both MSYS and Cygwin libraries.
(guile-system-extensions-path): prefer bindir to libdir on DLL systems
* test-suite/tests/foreign.test ("dll-name-match?"): new test category
("find-best-dll-from-matches"): new test category
("lib->msys"): new unit tests
Binding to AF_UNIX abstract sockets is not supported on Cygwin and
presumably MSYS as well.
* test-suite/tests/00-socket.text ("AF_UNIX abstract"): throw unsupported
on Cygwin and MSYS
Refactors a couple of the ports tests to catch errors in the test runner,
so that the test suite will print ERROR on failure.
* test-suite/tests/ports.test (pipe:write, pipe:read): modified
Since string-contains returns an integer on success, this test
was reporting unresolved instead of pass on success.
* test-suite/tests/version.test ("version reporting works"): fix
boolean check logic
Fixes a regression in ‘--without-threads’ builds introduced in
b8031fc965.
* libguile/finalizers.c (scm_i_is_finalizer_thread) [!SCM_USE_PTHREAD_THREADS]:
New function.
This fixes a problem on at least NetBSD.
* libguile/Makefile.am (libpath.h): move date -d argument before format
string.
Thanks to Thomas Klausner for reporting the problem and proposing the
fix.
Closes: 26121
Use scm_sendfile instead of read-write loop. This moves the work into
the kernel, improving performance. This implements Ludovic's suggestion
from https://debbugs.gnu.org/68504
* libguile/filesys.c (scm_copy_file2): Use scm_sendfile.
[rlb@defaultvalue.org: add NEWS]
struct timeval is a possible return value of getsockopt (e.g. SO_RCVTIMEO
and SO_SNDTIMEO), but it is not included in the scm_t_getsockopt_result
union, which may then be too small (and is on Debian amd64).
* libguile/socket.c: add struct timeval to scm_t_getsockopt union
[rlb@defaultvalue.org: adjust commit message; add NEWS]
Closes: 76907
The tests share a "log" for custom port events and didn't always
explicitly close the test ports, so the close might come later, during
another test. Change the tests to always close their ports immediately,
and clear the log after checking for expected "inter-test" events.
test-suite/tests/r6rs-ports.test: don't race with gc close in custom
port tests.
Without -ffat-lto-objects libguile.a ends up with no symbols, visible
via "nm t libguile.a".
cf. https://lintian.debian.org/tags/no-code-sections.html
* configure.ac: enable -ffat-lto-objects with -flto when available.
doc/ref/posix.texi: note setlocale raises a system-error when locale
doesn't exist
Thanks to Francesco P. Lovergine for suggesting the
addition (https://bugs.debian.org/1078681).
This is an amendment to 84bf840322.
The warning was only emitted for calling `environ', even if only reading
and no mutation occurred.
However, it is correct to read the environment in a multi-threaded
process. It is however unsafe to mutate it.
The same logic also applies to `putenv'.
* libguile/posix.c
(maybe_warn_about_environ_mutation): New private procedure ...
(scm_environ): ... called here when mutating the environment ...
(scm_putenv): ... and here.
* NEWS: Update.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Fixes <https://bugs.gnu.org/76343>.
Fixes a bug whereby “echo '(environ)' | guile” would wrongfully trigger
the multiple-thread warning.
* libguile/finalizers.c (finalizer_thread): New variable.
(finalization_thread_proc): Set it.
(scm_i_is_finalizer_thread): New function.
(run_finalization_thread): Clear FINALIZER_THREAD.
* libguile/finalizers.h (scm_i_is_finalizer_thread): New declaration.
* libguile/threads.c (scm_all_threads): Use it.
* NEWS: Update.
Reported-by: Simon Josefsson <simon@josefsson.org>
The tutorial was removed in commit f75c5849cd.
* doc/README: Remove reference to a tutorial that no longer exists
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* libguile/scm.h: BUILDING_LIBGUILE is not always defined. This is
signaled by -Werror=undef in code using libguile. This patch fixes
commit dc3a3a84f9
* NEWS: Update.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* module/ice-9/custom-ports.scm (make-custom-port): Code fails if
(fluid-ref %default-port-encoding) returns #f. In fact this was the
case why readline support on MSYS2 failed for guile 3.0.10, ref.
https://github.com/msys2/MSYS2-packages/issues/5079
But later used canonicalize-encoding is prepared to handle #f for
encoding. So allow encoding to also handle this case.
Co-authored-by: Ludovic Courtès <ludo@gnu.org>
The omission was discovered while cross-compiling LilyPond with mingw
11.0.1: without this addition, linking with libguile causes the error
undefined reference to `BCryptGenRandom'
* meta/Makefile.am (dependency_substitutions): Add ‘LIB_GETRANDOM’.
* meta/guile-3.0-uninstalled.pc.in (Libs.private): Likewise.
* meta/guile-3.0.pc.in (Libs.private): Likewise.
* NEWS: Update.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
The ~V is supposed to print ISO week number, not a week number. This
commit fixes that.
* module/srfi/srfi-19.scm (date-week-number-iso): New procedure taken
from the reference implementation.
(directives)<#\V>: Use it.
* test-suite/tests/srfi-19.test ("date->string ~V"): Add tests taken
from the reference test suite.
* doc/ref/srfi-modules.texi (SRFI-19 Date to string): Mention ISO-8601
in description for ~V.
Fixes <https://bugs.gnu.org/74841>.
Edited by lloda <lloda@sarc.name>.
RISC-V uses a0-a7 registers for argument passing. Float/double arguments
use f0-f7 first and continue in a0-a7 if needed.
Once registers are consumed, stack is used.
This commit changes how lightening passes arguments in order to allow
this behavior.
