The standalone egrep script has been deprecated by GNU grep.
'grep -E' is the suggested replacement.
* test-suite/tests/c-api.test (egrep): replace egrep with grep -E
* module/srfi/srfi-35.scm (condition): Use 'make-exception' instead of
'make-compound-condition', which is unbound in this module.
* test-suite/tests/srfi-35.test ("syntax")["compound condition,
hygienic macro expansion"]: New test.
* doc/ref/sxml.texi (7.21.4 Transforming SXML): Adds a mention to the
module to be imported for the procedures introduced in this section of
the documentation..
Co-authored-by: Ludovic Courtès <ludo@gnu.org>
Avoids spurious "error in finalization thread: Success" messages when
the finalization pipe gets closed.
* libguile/finalizers.c (finalization_thread_proc): Return when 'data.n'
is zero.
MinGW ACL-based permissions don't follow POSIX standard, so
'chmod' has unexpected behavior.
* test-suite/tests/ftw.test (mingw?): new define
("file system fold: EACCES"): skip test on MinGW
This is the same strategy as used for the 'accept4' bindings introduced
in 6e0965104c.
* libguile/posix.c (scm_pipe): Rename to...
(scm_pipe2): ... this. Add an optional 'flags' parameter and honor it.
(scm_pipe): Rewrite as a call to 'scm_pipe2'.
* libguile/posix.h (scm_pipe2): New declaration.
* test-suite/tests/posix.test ("pipe"): New tests.
* configure.ac: Look for 'pipe2'.
* NEWS: Update.
This makes the docstring attached to the curried function being defined
rather than the result of its application until a function that runs the
body is obtained, fixing
https://debbugs.gnu.org/cgi/bugreport.cgi?bug=50068
Locations are stored in tree-il records in "source vector" form, but
`location-string' was rendering these as <unknown-location>.
* module/system/base/message.scm (location-string): Support locations
passed as a file/line/column vector.
Fixes <https://bugs.gnu.org/55596>.
Reported by Hugo Nobrega <hugonobrega@ic.ufrj.br>
and Jack Hill <jackhill@jackhill.us>.
* libguile/posix.c (start_child): Close OUT only if it's greater than 2.
* test-suite/tests/posix.test ("system*")["exit code for nonexistent file"]
["https://bugs.gnu.org/55596"]: New tests.
Fixes <https://bugs.gnu.org/49223>.
Reported by Domagoj Stolfa <ds815@gmx.com>.
Backport of Guix commit b36267b1d96ac344d2b42c9822ce04b4c3117f85.
* guix/build/download.scm (tls-wrap): Retry up to 5 times when
'handshake' throws a non-fatal error.
The custom input/output port wrapping the TLS session record port would
introduce overhead, and it would also prevent its uses in a non-blocking
context--e.g., with Fibers. The port close mechanism added in GnuTLS
3.7.7 allows us to get rid of that wrapper.
Backported from Guix commit dd573ceea73295c7a872088ecd91e5f0fd74bf2b.
* web/client.scm (wrap-record-port-for-gnutls<3.7.7): New procedure,
with code formerly in 'tls-wrap'.
(tls-wrap): Check for 'set-session-record-port-close!' and use it when
available; otherwise call 'wrap-record-port-for-gnutls<3.7.7'.
This port is of limited use if it cannot be used reliably. Rather than
behaving as if the input has finished when it ends unexpectedly, instead
raise an exception.
* module/web/http.scm (make-chunked-input-port): Raise an exception on
premature termination.
(&chunked-input-ended-prematurely): New exception type.
(chunked-input-ended-prematurely-error?): New procedure.
* test-suite/tests/web-http.test (pass-if-named-exception): Rename to
pass-if-named-exception.
(pass-if-named-exception): New syntax.
("Exception on premature chunk end"): New test for this behaviour.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
The chunked transfer encoding specifies the chunked body ends with
CRLF. This is in addition to the CRLF at the end of the last chunk, so
there should be CRLF twice at the end of the chunked body:
https://datatracker.ietf.org/doc/html/rfc2616#section-3.6.1
* module/web/http.scm (make-chunked-input-port): Read two extra bytes at
the end of the chunked input.
(make-chunked-output-port): Write the missing \r\n when closing the
port.
* test-suite/tests/web-http.test (chunked encoding): Add missing \r\n to
test data.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
* module/web/http.scm (write-credentials): capitalize authorization
header scheme. The standard allows the scheme to be case-insensitive,
however most libraries out there expect the scheme to be capitalized,
which is what it is actually used in RFC
docs (e.g. https://datatracker.ietf.org/doc/html/rfc7617#section-2). Some
libraries even reject lowercase scheme making Guile incompatible.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
The code coverage function `coverage-data->lcov` has a documented
`modules` argument, however that was missing from the source. I have
added it so when supplied it only converts the coverage data for the
supplied modules. If not supplied it defaults the old behavour of
including all the modules currently loaded.
* module/system/vm/coverage.scm (coverage-data->lcov): Add #:modules
parameter and honor it.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
The current socket address constructors all assume, that there are no
null bytes in the socket path. This assumption does not hold in Linux,
which uses an initial null byte to demarcate abstract sockets and
ignores all further null bytes [1].
[1] https://www.man7.org/linux/man-pages/man7/unix.7.html
* libguile/sockets.c (scm_fill_sockaddr)[HAVE_UNIX_DOMAIN_SOCKETS]:
Use scm_to_locale_stringn to construct c_address.
Use memcpy instead of strcpy and calculate size directly instead of
using SUN_LEN.
(_scm_from_sockaddr): Copy the entire path up to the limits imposed by
addr_size.
* test-suite/tests/00-socket.test: ("make-socket-address"): Add case for
abstract unix sockets.
("AF_UNIX/SOCK_STREAM"): Add abstract socket versions of bind, listen,
connect and accept.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
NetBSD and pkgsrc have been using an empty vendor string since the
mid-'90s, such as x86_64--netbsd. pkgsrc has been carrying around a
workaround just the guile build for a long time. (Before that,
NetBSD omitted the vendor altogether, so if x86_64 existed then it
might have been `x86_64-netbsd', but that caused more problems.)
This change makes Guile accept an empty vendor string so workarounds
are no longer necessary.
* module/system/base/target.scm (validate-target): Allow empty vendor
string in GNU target triplets.
* test-suite/tests/cross-compilation.test ("cross-compilation"): Add
tests for "x86_64--netbsd".
Co-authored-by: Ludovic Courtès <ludo@gnu.org>
Guile (3.0.8) reports a compilation error when cond-expand tries to
check existence of a missing library:
scheme@(guile-user)> (define-library (test)
(cond-expand
((library (scheme sort))
(import (scheme sort)))))
While compiling expression:
no code for module (scheme sort)
It looks like bug #40252 was not fully eliminated.
Also, (library ...) cannot handle module names like (srfi 1), though
(import (srfi 1)) works fine. For example, this code fails:
scheme@(guile-user)> (define-library (test)
(cond-expand
((library (srfi 1))
(import (srfi 1)))))
While compiling expression:
In procedure symbol->string: Wrong type argument in position 1
(expecting symbol): 1
There are probably other cases when (library ...) and (import ...) does
not work identically: (library ...) uses resolve-interface while
(import ...) uses resolve-r6rs-interface.
This patch fixes both issues.
* module/ice-9/r7rs-libraries.scm (define-library): Replace
'resolve-interface' call by 'resolve-r6rs-interface', wrapped in
'cond-expand'.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Don't close the test port's file descriptor because the port still has a
reference to it, and could still close it at any time when finally
garbage collected. This did soemetimes break subsequent tests.
Bug: https://debbugs.gnu.org/43521
Fixes synchronization issues observed on slow or loaded machines, where
client connection attempts would fail with ECONNREFUSED:
https://issues.guix.gnu.org/54348
* test-suite/tests/web-server.test ("server is listening"): New test.
Before the change ./configure incorrectly enabled -flto on toolchains
that support -flto on compiler side but don't support -flto on linker
side. This caused incorrect type size detection on nixpkgs' Darwin:
configure:54594: checking size of size_t
configure:54600: clang -std=gnu11 -o conftest -g -O2 -flto conftest.c >&5
ld: warning: ignoring file /private/tmp/nix-build-guile-3.0.8.drv-0/conftest-00e93d.o,
building for macOS-x86_64 but attempting to link with file built
for unknown-unsupported file format ( 0xDE 0xC0 0x17 0x0B 0x00 0x00 0x00 0x00 0x14 0x00 0x00 0x00 0x80 0x1A 0x00 0x00 )
Undefined symbols for architecture x86_64:
"_main", referenced from:
implicit entry/start for main executable
ld: symbol(s) not found for architecture x86_64
clang-11: error: linker command failed with exit code 1 (use -v to see invocation)
Taken from https://github.com/NixOS/nixpkgs/pull/160051#issuecomment-1046105041
The change makes sure -flto support tests basic support of just for
object file generation but for linker as well.
* configure.ac: use AC_LINK_IFELSE instead of AC_COMPILE_IFELSE.
Signed-off-by: Ludovic Courtès <ludo@gnu.org>
Commit 54bbe0b284 inadvertently led
psyntax to dismiss source location info for data returned by read hash
extensions, because read hash extensions return plain data with
associated source properties, even when called from 'read-syntax'.
This change reverts part of this commit to restore that behavior.
Fixes <https://issues.guix.gnu.org/54003>.
* module/ice-9/psyntax.scm (datum-sourcev): New procedure.
(source-annotation): Fall back to 'datum-sourcev'.
* module/ice-9/psyntax-pp.scm: Regenerate.
* test-suite/tests/compiler.test ("psyntax")["syntax-source with
read-hash-extend"]: New test.
Fixes <https://issues.guix.gnu.org/54198>.
As noted in the comment at the top, 'SIZEOF_TYPE' must be used instead
of 'sizeof (TYPE)' to support cross-compilation.
The regression was introduced in
5e5afde06f but only became apparent with
717e787da6.
* libguile/gen-scmconfig.c (main): Replace uses of 'sizeof' by
references to the SIZEOF_* macros.
* configure.ac: Add 'AC_CHECK_SIZEOF' call for 'intmax_t'.
