To fix this properly, we'd need to configure timeout for the server and
the client because waiting for compilation takes time.
Until we figure out how to do it, let me skip this.
... because it conflicts with test/ruby/test_m17n.rb.
An exception `incompatible character encodings: UTF-8 and UTF-16BE`
occurs when:
* a non-existence relative path is added to $LOAD_PATH,
* ASCII-incompatible encoding is set to default_external, and
* some file is loaded.
```
$LOAD_PATH << "no_existing_dir"
Encoding.default_external = Encoding::UTF_16BE
load "dummy.rb" #=> incompatible character encodings: UTF-8 and UTF-16BE
```
This issue can be actually observed by a combination of out-of-place
build and the following command:
make test-all TESTS="json ruby/m17n -n test_object_inspect_external"
http://ci.rvm.jp/logfiles/brlog.trunk-test-random.20200322-221411
ASCII-incompatible default external encoding assumes that the cwd is the
encoding, and it is attempted to beconcatenated with a non-existence
relative LOAD_PATH UTF-8 string, which causes the exception.
This changeset avoids a relative path.
This freezes the clone even if the receiver is not frozen. It
is only for consistency with freeze: false not freezing the clone
even if the receiver is frozen.
Because Object#clone is now partially implemented in Ruby and
not fully implemented in C, freeze: nil must be supported to
provide the default behavior of only freezing the clone if the
receiver is frozen.
This requires modifying delegate and set, to set freeze: nil
instead of freeze: true as the keyword parameter for
initialize_clone. Those are the two libraries in stdlib that
override initialize_clone.
Implements [Feature #16175]
http://ci.rvm.jp/logfiles/brlog.trunk-test-random.20200322-221411
```
I, [2020-03-22T22:15:50.761950 #23076] INFO -- : 1) Error:
I, [2020-03-22T22:15:50.761963 #23076] INFO -- : TestM17N#test_object_inspect_external:
I, [2020-03-22T22:15:50.761974 #23076] INFO -- : Encoding::CompatibilityError: incompatible character encodings: UTF-8 and UTF-16BE
```
The test fails randomly on the CI of OpenCSW SPARC Solaris 10:
https://rubyci.org/logs/rubyci.s3.amazonaws.com/unstable10s/ruby-master/log/20200321T091909Z.fail.html.gz
```
1) Failure:
TestNetHTTP_v1_2_chunked#test_timeout_during_HTTP_session [/export/home/rubyci/unstable10s/tmp/build/20200321T091909Z/ruby/test/net/http/test_http.rb:575]:
[Net::ReadTimeout] exception expected, not #<Net::OpenTimeout: execution expired>.
```
The environment uses RUBY_TEST_SUBPROCESS_TIMEOUT_SCALE=20, so the open
timeout 0.1 sec. means 2.0 sec. for the environment, but it seems too
strict because the environment is painfully slow.
"test_close_after_socket_close" checks if ssl.close is no-op even after
the wrapped socket is closed. The test itself is fair, but the other
endpoint that is reading the SSL connection may fail with SSLError:
"SSL_read: unexpected eof while reading" in some environments:
https://github.com/ruby/ruby/actions/runs/60085389 (MinGW)
https://rubyci.org/logs/rubyci.s3.amazonaws.com/android28-x86_64/ruby-master/log/20200321T034442Z.fail.html.gz
```
1) Failure:
OpenSSL::TestSSL#test_close_after_socket_close [D:/a/ruby/ruby/src/test/openssl/utils.rb:299]:
exceptions on 1 threads:
SSL_read: unexpected eof while reading
```
This changeset rescues and ignores the SSLError in the test.
Previously, passing a keyword splat to a method always allocated
a hash on the caller side, and accepting arbitrary keywords in
a method allocated a separate hash on the callee side. Passing
explicit keywords to a method that accepted a keyword splat
did not allocate a hash on the caller side, but resulted in two
hashes allocated on the callee side.
This commit makes passing a single keyword splat to a method not
allocate a hash on the caller side. Passing multiple keyword
splats or a mix of explicit keywords and a keyword splat still
generates a hash on the caller side. On the callee side,
if arbitrary keywords are not accepted, it does not allocate a
hash. If arbitrary keywords are accepted, it will allocate a
hash, but this commit uses a callinfo flag to indicate whether
the caller already allocated a hash, and if so, the callee can
use the passed hash without duplicating it. So this commit
should make it so that a maximum of a single hash is allocated
during method calls.
To set the callinfo flag appropriately, method call argument
compilation checks if only a single keyword splat is given.
If only one keyword splat is given, the VM_CALL_KW_SPLAT_MUT
callinfo flag is not set, since in that case the keyword
splat is passed directly and not mutable. If more than one
splat is used, a new hash needs to be generated on the caller
side, and in that case the callinfo flag is set, indicating
the keyword splat is mutable by the callee.
In compile_hash, used for both hash and keyword argument
compilation, if compiling keyword arguments and only a
single keyword splat is used, pass the argument directly.
On the caller side, in vm_args.c, the callinfo flag needs to
be recognized and handled. Because the keyword splat
argument may not be a hash, it needs to be converted to a
hash first if not. Then, unless the callinfo flag is set,
the hash needs to be duplicated. The temporary copy of the
callinfo flag, kw_flag, is updated if a hash was duplicated,
to prevent the need to duplicate it again. If we are
converting to a hash or duplicating a hash, we need to update
the argument array, which can including duplicating the
positional splat array if one was passed. CALLER_SETUP_ARG
and a couple other places needs to be modified to handle
similar issues for other types of calls.
This includes fairly comprehensive tests for different ways
keywords are handled internally, checking that you get equal
results but that keyword splats on the caller side result in
distinct objects for keyword rest parameters.
Included are benchmarks for keyword argument calls.
Brief results when compiled without optimization:
def kw(a: 1) a end
def kws(**kw) kw end
h = {a: 1}
kw(a: 1) # about same
kw(**h) # 2.37x faster
kws(a: 1) # 1.30x faster
kws(**h) # 2.19x faster
kw(a: 1, **h) # 1.03x slower
kw(**h, **h) # about same
kws(a: 1, **h) # 1.16x faster
kws(**h, **h) # 1.14x faster
Previously, method call keyword splats and hash keyword splats
were compiled exactly the same. This is because parse-wise, they
operate on indentical nodes when it comes to compiling the **{}.
Fix this by using an ugly hack of temporarily modifying the
nd_brace flag in the method call keyword splat case. Inside
compile_hash, only optimize the **{} case for hashes where the
nd_brace flag has been modified to reflect we are in the method
call keyword splat case and it is safe to do so.
Since compile_keyword_args is only called in one place, move the
keyword_node_p call out of that method to the single caller to
avoid duplicating the code.
As a semantics, Hash#each yields a 2-element array (pairs of keys and
values). So, `{ a: 1 }.each(&->(k, v) { })` should raise an exception
due to lambda's arity check.
However, the optimization that avoids Array allocation by using
rb_yield_values for blocks whose arity is more than 1 (introduced at
b9d2960337 and some commits), seemed to
overlook the lambda case, and wrongly allowed the code above to work.
This change experimentally attempts to make it strict; now the code
above raises an ArgumentError. This is an incompatible change; if the
compatibility issue is bigger than our expectation, it may be reverted
(until Ruby 3.0 release).
[Bug #12706]
* Let Net::HTTP.get take request headers
* Add more test cases for no header usages
* Add examples with request headers
* Add a NEWS entry [ci skip]
[Feature #16686]
To invalidate cached method entry, existing method entry (ment)
is marked as invalidated and replace with copied ment. However,
complemented method entry (method entries in Module) should not
be set to Module's m_tbl.
[Bug #16669]
Import current master (2c43241dc0ed) of ruby/openssl.git.
Below are the commits that were made since the last batch at commit
b99775b163 (ruby/openssl.git commit f49e7110ca1e). Note that some of
them have been applied already.
----------------------------------------------------------------
Benoit Daloze (1):
Remove redundant and ignored workflow file
DBL-Lee (1):
add support for SHA512_256/SHA512_224
Hiroshi SHIBATA (2):
Guard for OpenSSL::PKey::EC::Group::Error with unsupported platforms
Fixed inconsistency directory structure with ruby/ruby repo
Jeremy Evans (2):
Fix keyword argument separation issues in OpenSSL::SSL::SSLSocket#sys{read,write}_nonblock
Remove taint support
Kazuki Yamaguchi (26):
config: support .include directive
random: make OpenSSL::Random.pseudo_bytes alias of .random_bytes
extconf.rb: get rid of -Werror=deprecated-declarations
test/openssl/test_ssl: skip test_fallback_scsv if necessary
ts: simplify OpenSSL::Timestamp::Request#algorithm
History.md: add missing references to GitHub issues
config: deprecate OpenSSL::Config#add_value and #[]=
test/openssl/test_ssl: remove sleep from test_finished_messages
test/openssl/test_ssl: fix random failure in SSLSocket.open test
test/openssl/test_ssl: avoid explicitly-sized private keys
test/openssl/test_ssl: remove commented-out test case
test/openssl/test_ssl: allow kRSA tests to fail
ssl: avoid declarations after statements
engine: revert OpenSSL::Engine.load changes for cloudhsm
engine: remove really outdated static engines
engine: do not check for ENGINE_load_builtin_engines()
engine: fix guards for 'dynamic' and 'cryptodev' engines
lib/openssl.rb: require openssl/version.rb
x509: add error code and verify flags constants
ssl: set verify error code in the case of verify_hostname failure
.github/workflows: merge CI jobs into a single workflow
.github/workflows: test against different OpenSSL versions
.travis.yml: fully migrate to GitHub Actions
ssl: suppress test failure with SSLContext#add_certificate_chain_file
ssl: remove test case test_puts_meta from test_pair
Revert "Use version.rb in gemspec"
MSP-Greg (2):
.travis.yml - remove 2.3/1.0.2, 2.5/1.1.1, head/1.0.2
Use version.rb in gemspec
Samuel Williams (1):
Restore compatibility with older versions of Ruby.
Yusuke Endoh (1):
Make OpenSSL::OSSL#test_memcmp_timing robust
Previously, if an object has a singleton class, and you call
Object#method on the object, the resulting string would include
the object's singleton class, even though the method was not
defined in the singleton class.
Change this so the we only show the singleton class if the method
is defined in the singleton class.
Fixes [Bug #15608]
When providing a single array to a block that takes a splat, pass the
array as one argument of the splat instead of as the splat itself,
even if the block also accepts keyword arguments. Previously, this
behavior was only used for blocks that did not accept keywords.
Implements [Feature#16166]
ar_table can be converted to st_table just after `ar_do_hash()`
function which calls `#hash` method. We need to check
the representation to detect this mutation.
[Bug #16676]
Previously, it would be an infinite loop if passed a non-prime
integer.
Also, Prime.include? should also provide similar results to
Module#include? if passed a Module, so handle that.
For consistency with Enumerable#include?, return false if passed
other object types.
Fixes Ruby Bug 10167.
https://github.com/ruby/prime/commit/55dda6aa7f
`GC.start(full_mark: false)` can run full GC because of last
GC status. Just after major GC, the possibility to run major GC
next time is too small (not a zero, but too small possibility).
Surprisingly (at least for me), `server[:DocumentRootOptions]`
on Webrick is global information and it affect the result of
test_short_filename. Random order test fails because of global
information change. I doubt it is bug, but to fix random order
test, I restore the value.
test_readline:
HISTORY should be empty.
test_using_quoting_detection_proc:
test_using_quoting_detection_proc_with_multibyte_input:
Readline.completer_quote_characters= and
Readline.completer_word_break_characters= doesn't accept nil,
so skip if previous values are nil.
When calling Module#include, if the receiver is a module,
walk the subclasses list and include the argument module in each
iclass.
This does not affect Module#prepend, as fixing that is significantly
more involved.
Fixes [Bug #9573]
In ruby/ruby, the tests run on both readline & reline by creating four test classes:
```
TestReadline
TestReadlineHistory
TestRelineAsReadline
TestRelineAsReadlineHistory
```
Reline inports the test files and uses them in its CI. Adding the ENV control allows it to only run the `TestRelineAsReadline` classes.
This patch contains several ideas:
(1) Disposable inline method cache (IMC) for race-free inline method cache
* Making call-cache (CC) as a RVALUE (GC target object) and allocate new
CC on cache miss.
* This technique allows race-free access from parallel processing
elements like RCU.
(2) Introduce per-Class method cache (pCMC)
* Instead of fixed-size global method cache (GMC), pCMC allows flexible
cache size.
* Caching CCs reduces CC allocation and allow sharing CC's fast-path
between same call-info (CI) call-sites.
(3) Invalidate an inline method cache by invalidating corresponding method
entries (MEs)
* Instead of using class serials, we set "invalidated" flag for method
entry itself to represent cache invalidation.
* Compare with using class serials, the impact of method modification
(add/overwrite/delete) is small.
* Updating class serials invalidate all method caches of the class and
sub-classes.
* Proposed approach only invalidate the method cache of only one ME.
See [Feature #16614] for more details.
Now, rb_call_info contains how to call the method with tuple of
(mid, orig_argc, flags, kwarg). Most of cases, kwarg == NULL and
mid+argc+flags only requires 64bits. So this patch packed
rb_call_info to VALUE (1 word) on such cases. If we can not
represent it in VALUE, then use imemo_callinfo which contains
conventional callinfo (rb_callinfo, renamed from rb_call_info).
iseq->body->ci_kw_size is removed because all of callinfo is VALUE
size (packed ci or a pointer to imemo_callinfo).
To access ci information, we need to use these functions:
vm_ci_mid(ci), _flag(ci), _argc(ci), _kwarg(ci).
struct rb_call_info_kw_arg is renamed to rb_callinfo_kwarg.
rb_funcallv_with_cc() and rb_method_basic_definition_p_with_cc()
is temporary removed because cd->ci should be marked.
Stop the special treatment of invalid hashAlgorithm of the message
imprint. Those invalid values can only appear after the object is
instantiated, before the user sets an actual message digest algorithm.
OpenSSL::Timestamp::TokenInfo#algorithm already does the same.
Also, remove the test case "test_create_request" since it does not make
much sense. Those fields are to be set by the user after creation of
the object and checking the initial value is pointless.
Fixes: https://github.com/ruby/openssl/issues/335https://github.com/ruby/openssl/commit/890a6476fa
It fails due to "error:140AB18F:SSL routines:SSL_CTX_use_certificate:ee
key too small". This is a tentative measurement to avoid the failure.
https://rubyci.org/logs/rubyci.s3.amazonaws.com/debian/ruby-master/log/20200216T093010Z.fail.html.gz
test/openssl/fixture/chain/server.key should be longer.
It should be documented how to create the files.
BTW, it would be a good idea to dynamically create a key during test
instead of fixed files.
After the encoding index instance variable is removed when all
instance variables are removed in `obj_free`, then `rb_str_free`
causes uninitialized instance variable warning and nil-to-integer
conversion exception. Both cases result in object allocation
during GC, and crashes.
The previous version of the test method used a symbol, ":abcdefg" to complete
but longer symbols that can be completed are defined by other test methods of
other libs.
Since 9d9aea7fe5, generic instance
variables need `iv_index_tbl` in the object's class. As hidden
objects, however, have no class, access to the variables causes a
segfault. Get rid of that segfault by raising an exception, for
the time being.
This behavior was deprecated in 2.7 and scheduled to be removed
in 3.0.
Calling yield in a class definition outside a method is now a
SyntaxError instead of a LocalJumpError, as well.
In 2.6, SimpleDelegator would always use the target `respond_to?`
In 2.7.0 it doesn't if the target does not inherit from Object.
This breaks compatibility for delegated objects that inherit
from BasicObject and redefine `respond_to?`.
https://rubyci.org/logs/rubyci.s3.amazonaws.com/ubuntu1604/ruby-master/log/20200203T003005Z.log.html.gz
```
/home/hsbt/chkbuild/tmp/build/20200203T003005Z/ruby/test/rubygems/test_gem_remote_fetcher.rb:485: warning: method redefined; discarding old test_fetch_path_socket_error
/home/hsbt/chkbuild/tmp/build/20200203T003005Z/ruby/test/rubygems/test_gem_remote_fetcher.rb:167: warning: previous definition of test_fetch_path_socket_error was here
```
test_rand.rb calls srand() several times, however it change global
rand-sequence (and --seeds doesn't have meaning). This patch makes
such tests run in other processes.
ruby/test_time_tz.rb (not sure only this file affects) changes TZ
and it seems to change internal state. This internal state change
fails test_2038 and test_timegm on 2nd time execution.
At this time I have no idea how to fix this issue, so I skips these
tests on 2nd trial.
You can try this failure with the following command without this patch.
$ make test-all TESTS='--repeat-count=2 ruby/require ruby/time ruby/time_tz'
Mysterious error:
`remove_method(:foo) if method_defined?(:foo)` raise an exception
`method `foo' not defined in #<Class:#<TestProc:0x000055d12ff154e0>>`
This patch rename the method name foo to foo_arity to solve it.
With the removal of the splatted argument when using an empty
keyword splat, the autosplat code considered an empty keyword
splat the same as no argument at all. However, that results
in autosplat behavior changing dependent on the content of
the splatted hash, which is not what anyone would expect or
want. This change always skips an autosplat if keywords were
provided.
Fixes [Bug #16560]
Musl libc has this function as a tiny wrapper of fchmodat(3posix). On
the other hand Linux kernel does not support changing modes of a symlink.
The operation always fails with EOPNOTSUPP. This fchmodat behaviour is
defined in POSIX. We have to take care of such exceptions.
Musl is (of course) not glibc. Its confstr(3) does not understand
_CS_GNU_LIBC_VERSION. That's fair. Problem is, its unistd.h has that
constant defined for unknown reason. We cannot blindly say the libc is
glibc by looking at it. Instead we have to kick it, then see if it
quacks like a duck.
See https://git.musl-libc.org/cgit/musl/tree/include/unistd.h
The same as https://github.com/ruby/ruby/pull/2686, but for musl libc.
Musl is not named as libc.so.6 so the `ldd` hack implemented some lines
below does not work.
Keeping empty keyword splats for ruby2_keywords methods was
necessary in 2.7 to prevent the final positional hash being
treated as keywords. Now that keyword argument separation
has been committed, the final positional hash is never
treated as keywords, so there is no need to keep empty
keyword splats when using ruby2_keywords.
According to https://github.com/ruby/openssl/pull/60,
> Currently an user who wants to do the hostname verification needs to
call SSLSocket#post_connection_check explicitly after the TLS connection
is established.
if an user who wants to skip the hostname verification,
SSLSocket#post_connection_check doesn't need to be called
https://bugs.ruby-lang.org/issues/16555
This causes problems because the hash is passed to a block not
accepting keywords. Because the hash is empty and keyword flagged,
it is removed before calling the block. This doesn't cause an
ArgumentError because it is a block and not a lambda. Just like
any other block not passed required arguments, arguments not
passed are set to nil.
Issues like this are a strong reason not to have ruby2_keywords
by default.
Fixes [Bug #16519]
Matt Muller
aycabta
Takashi Kokubun
Nobuyoshi Nakada
Yusuke Endoh
David Rodríguez
Jeremy Evans
Jean Boussier
Koichi Sasada
Kazuki Yamaguchi
Charles Oliver Nutter
Kazuki Tsujimoto
Kazuhiro NISHIYAMA
MSP-Greg
git
Hiroshi SHIBATA
卜部昌平
S-H-GAMELINKS
Nobuhiro IMAI
John Hawthorn
Lars Kanis
Seiei Miyagi
Tanaka Akira
Benoit Daloze
Yuta Iwama