In an effort to minimize build issues on non x64 platforms, we can
decide at build time to not build the bulk of YJIT. This should fix
obscure build errors like this one on riscv64:
yjit_asm.c:137:(.text+0x3fa): relocation truncated to fit: R_RISCV_PCREL_HI20 against `alloc_exec_mem'
We also don't need to bulid YJIT on `--disable-jit-support` builds.
One wrinkle to this is that the YJIT Ruby module will not be defined
when YJIT is stripped from the build. I think that's a fair change as
it's only meant to be used for YJIT development.
`duphash` showed up in the top-20 most frequent exit ops for @jhawthorn's benchmark that renders github.com/about
The implementation was almost exactly the same as `duparray`
Co-authored-by: John Hawthorn <john@hawthorn.email>
Co-authored-by: John Hawthorn <john@hawthorn.email>
Otherwise, the hint is printed whenever fork is called.
http://rubyci.s3.amazonaws.com/debian9/ruby-master/log/20211025T093004Z.log.html.gz
```
[20244/21156] TestThread#test_fork_while_lockedyou may want to add tests for following insns, when you have a chance: checkmatch
you may want to add tests for following insns, when you have a chance: checkmatch
you may want to add tests for following insns, when you have a chance: checkmatch
= 0.19 s
```
We require things inline or autoload them for efficiency and to avoid as
much as possible conflicting with user's choice of gems.
But I removed the require and observed no deadlocks when activating
gems.
https://github.com/rubygems/rubygems/commit/7d6333e842
Some crucial information to ease maintainers work, like the advice of
upgrading rubygems and bundler, was one step away from the issue
template, making it easier for some users to miss.
Now all relevant information is written directly in the bug report
template.
We can skip most stuff in `Gem::RemoteFetcher#download`, and use
`Gem::RemoteFetcher#update_cache_path` directly.
This has the benefit of allowing us to remove some workarounds to
support several rubygems versions, but also allows us to pass the target
folder where the gem should be downloaded directly and skip the logic
inside `Gem::RemoteFetcher#download` to infer the cache path. This will
be useful later to fix some issues with the `global_gem_cache` feature
flag.
https://github.com/rubygems/rubygems/commit/8fe74a77e4
These methods rescue a constant defined by `rubygems/remote_fetcher`,
so they should technically require it.
The require is provided by `gem_remote_fetcher` anyways but I was
running a unit spec that stubs that method, so I was getting an
undefined constant error hiding another error.
https://github.com/rubygems/rubygems/commit/8bedae4034
Extract final cache path to a variable and pass that to `download_gem`.
It actually fits better the parameters documentation since it's the
final directory where the downloaded gem will be placed.
https://github.com/rubygems/rubygems/commit/1429db6a04
* process.c: Add Process._fork
This API is supposed for application monitoring libraries to hook fork
event.
[Feature #17795]
Co-authored-by: Nobuyoshi Nakada <nobu@ruby-lang.org>
http://rubyci.s3.amazonaws.com/centos7/ruby-master/log/20211025T093004Z.log.html.gz
```
[ 4896/21159] TestRubyVM#test_keep_script_lines(none):3: warning: assigned but unused variable - b
(none):6: warning: assigned but unused variable - c
(none):1: warning: assigned but unused variable - a
<compiled>:3: warning: assigned but unused variable - b
(none):3: warning: assigned but unused variable - b
(none):6: warning: assigned but unused variable - c
(none):1: warning: assigned but unused variable - a
<compiled>:3: warning: assigned but unused variable - b
```
http://rubyci.s3.amazonaws.com/centos7/ruby-master/log/20211025T093004Z.log.html.gz
```
/home/chkbuild/chkbuild/tmp/build/20211025T093004Z/ruby/test/ruby/test_refinement.rb:777: warning: Refinement#include is deprecated and will be removed in Ruby 3.2
/home/chkbuild/chkbuild/tmp/build/20211025T093004Z/ruby/test/ruby/test_refinement.rb:840: warning: Refinement#prepend is deprecated and will be removed in Ruby 3.2
/home/chkbuild/chkbuild/tmp/build/20211025T093004Z/ruby/test/ruby/test_refinement.rb:2620: warning: Refinement#include is deprecated and will be removed in Ruby 3.2
```
When `install_with_build_args` was added in
be96283985,
there were two versions of the method: the default version in the base class that still
used the locking `with_build_args`, and an override in the `Future`
class (for Rubygems 2.0 and up) that yielded without calling
`with_build_args`.
The `with_build_args` version of the method was removed in
8a5b71e3e8
while removing a bunch of the old Rubygems compatibility code.
This commit removes `with_build_args`, since it no longer appears to be
used (the build args are passed as a keyword argument to
`spec.source.install` instead, since
be96283985).
The commit also removes `install_with_build_args` and the conditional
around it, since the method wasn't doing anything different than
`install`, and it had a comment that was no longer accurate.
https://github.com/rubygems/rubygems/commit/ba543a60eb
BN_pseudo_rand() and BN_pseudo_rand_range() are deprecated in
OpenSSL 3.0. Since they are identical to their non-'pseudo' version
anyway, let's make them alias.
https://github.com/ruby/openssl/commit/2d34e85ddf
It converts the internal representation of the point object to the
affine coordinate system. However, it had no real use case because the
difference in the internal representation has not been visible from
Ruby/OpenSSL at all.
EC_POINT_make_affine() is marked as deprecated in OpenSSL 3.0.
https://github.com/ruby/openssl/commit/e2cc81fef7
OpenSSL 3.0 renamed EVP_MD_CTX_pkey_ctx() to include "get" in the
function name. Adjust compatibility macro so that we can use the new
function name for all OpenSSL 1.0.2-3.0.
https://github.com/ruby/openssl/commit/c106d888c6
The function was renamed in OpenSSL 3.0 due to the change of the
lifetime of EVP_MD objects. They are no longer necessarily statically
allocated and can be reference-counted -- when an EVP_MD_CTX is free'd,
the associated EVP_MD can also become inaccessible.
Currently Ruby/OpenSSL only handles builtin algorithms, so no special
handling is needed except for adapting to the rename.
https://github.com/ruby/openssl/commit/0a253027e6
Use SSL_get_rbio() instead of SSL_get_fd(). SSL_get_fd() internally
calls SSL_get_rbio() and it's enough for our purpose.
In OpenSSL 3.0, SSL_get_fd() leaves an entry in the OpenSSL error queue
if BIO has not been set up yet, and we would have to clean it up.
https://github.com/ruby/openssl/commit/e95ee24867
Alan Wu
Ian C. Anderson
Yusuke Endoh
David Rodriguez
Jean Boussier
Daniel Colson
Koichi Sasada
Tanaka Akira
180909
TSUYUSATO Kitsune
Kazuhiro NISHIYAMA
Nobuyoshi Nakada
git
Kazuki Yamaguchi