Age | Commit message (Collapse) | Author |
|
As `RbConfig.expand` modifies the argument and involved `CONFIG`
values, its values should be mutable.
|
|
|
|
for consistency
|
|
variable of make instead of d059714746 as requested.
You can disable this like `make MJIT_WITHOUT_TABS=false` or
`export MJIT_WITHOUT_TABS=false`.
|
|
in debugflags, as requested by nobu
I actually wanted to check debugflags directly rather than MJIT_DEBUGFLAGS
because debugflags is always used but MJIT_DEBUGFLAGS may not be always used.
But I couldn't find a better way to check debugflags.
|
|
rb_class_of is indented with tabs
|
|
This didn't exist in my machine, but it may exist in some CI
environments.
|
|
|
|
It's to be fixed properly later. But this should be able to avoid the failure.
https://travis-ci.org/github/ruby/ruby/jobs/665580361
https://rubyci.org/logs/rubyci.s3.amazonaws.com/solaris10-gcc/ruby-master/log/20200322T190003Z.fail.html.gz
|
|
This is necessary to avoid converting a hard tab to just 1 space in
preprocessor to generate rb_mjit_header.h, which is helpful when using
gdb or perf report.
See also: [Misc #16112]
This reverts commit 91acdd17c4b4bb69a8fa3ada46e09dad46b9362e.
Fixed permission failure on Travis, encoding, and added SKIPPED_FILES.
Notes:
Merged-By: k0kubun <takashikkbn@gmail.com>
|
|
This reverts commit 57119dd561418c917b885db5f5af7f129a96d1ec.
Temporarily reverting for Travis failures
|
|
I can't live without this when using gdb or perf report.
See also: [Misc #16112]
|
|
Notes:
Merged: https://github.com/ruby/ruby/pull/2972
|
|
It seems the issue was fixed by Travis.
See https://travis-ci.community/t/6719/5 .
Notes:
Merged: https://github.com/ruby/ruby/pull/2970
|
|
https://github.com/rubygems/rubygems/pull/3166
|
|
|
|
|
|
OpenBSD RubyCI has failed with SEGV since 4bcd5981e80d3e1852c8723741a0069779464128.
https://rubyci.org/logs/rubyci.s3.amazonaws.com/openbsd-current/ruby-master/log/20200312T223005Z.fail.html.gz
This was because `status->cc_entries` could be stale after `realloc` call
for inlined iseqs.
|
|
```
$ benchmark-driver benchmark.yml -v --rbenv 'before --jit;after --jit' --repeat-count=12 --output=all
before --jit: ruby 2.8.0dev (2020-03-11T07:43:12Z master e89ebdcb87) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-03-11T07:54:18Z master 143776a0da) +JIT [x86_64-linux]
Calculating -------------------------------------
before --jit after --jit
Optcarrot Lan_Master.nes 73.86976729561439 77.20184819316513 fps
74.46997176460742 78.43493030231805
77.59686308754307 78.55714131655935
78.53693921126656 79.08984255596820
80.10158944910573 79.17751731838183
80.12254974411167 79.60853122429181
80.28678655204945 79.74674066871896
80.38690681095379 79.90624544440300
80.79223498756919 80.57881084206193
80.82857188422419 80.70677614429169
81.06447745878245 81.03868541295149
81.21620802278490 82.16354660940607
```
|
|
According to ko1, `cd->cc != cc` was for GC.compact guard.
As we pin cc by rb_gc_mark(), we don't need the check.
```
$ benchmark-driver benchmark.yml -v --rbenv 'before --jit;after --jit' --repeat-count=12 --output=all
before --jit: ruby 2.8.0dev (2020-03-11T05:36:48Z master da6948753e) +JIT [x86_64-linux]
after --jit: ruby 2.8.0dev (2020-03-11T06:26:34Z master 36b20b8b4a) +JIT [x86_64-linux]
Calculating -------------------------------------
before --jit after --jit
Optcarrot Lan_Master.nes 74.03480698689405 71.63404803273507 fps
74.15085286586992 73.43923328104295
75.51738277744781 75.75465268365384
76.24922600109410 76.74071607861318
76.45513422802325 77.47521029238116
76.86617230739330 78.14759496269018
77.71509137131933 79.14051571125866
77.72839157096146 79.35884822673313
78.25218904561633 79.92538876408051
78.72521071333249 79.98075556706726
78.79950460165091 80.51747831497875
79.43884960720381 80.97973166525254
```
|
|
Now ruby/openssl has a different directory structure. The test suite is
moved from /test/test_* to /test/openssl/test_*.
|
|
Fixing https://github.com/ruby/ruby/blob/4bcd5981e80d3e1852c8723741a0069779464128/mjit.c#L338
should be the right solution for this. We may not be able to free the cc immediately.
Plus, we're not copying cc but just holding references to be marked. cc
should be GC-ed once jit_unit is freed.
|
|
jit_unit to avoid marking wrong cc entries when inlined iseq is compiled
multiple times, resolving the TODO added by daf7c48d88.
This obviates pseudo jit_unit in inlined iseq introduced by 7ec2359374
and fixes memory leak of the adhoc unit.
|
|
Retry with 90de118a75921a08497b41de2950971073b3c924
[Bug #13962][ruby-core:83074]
|
|
|
|
* Fix incorrect calls to `Gem.ensure_gem_subdirectories`
This method doesn't take keyword args.
* Remove stuff no longer necessary
Now `Gem.ensure_gem_subdirectories` is doing its job, so some stuff is
no longer needed.
* Use the proper method for default gems
* Respect DESTDIR when creating rubygems folder layout
* Use `Gem.default_specifications_dir`
Notes:
Merged-By: hsbt <hsbt@ruby-lang.org>
|
|
There is a test directory named test/-ext-. Because this directry
starts with a hyphen, we have to cheat test/unit in order for it to
run the tests underneath. TESTS='-- -ext' worked for a long time.
Let's not break that maneuver.
|
|
Call GC.verify_internal_consistency only for Gem related test
on i386-solaris.
|
|
|
|
Check heap consistency after each test to debug.
https://rubyci.org/logs/rubyci.s3.amazonaws.com/unstable11x/ruby-master/log/20200303T012406Z.fail.html.gz
|
|
assert_nothing_raised doesn't show the backtrace if an exception
is thrown in a block. This patch shows this backtrace.
|
|
Now --test-order=random is simply ignored. This patch respect
this option.
Notes:
Merged: https://github.com/ruby/ruby/pull/2932
|
|
The missing `\` in PR #2922 causes the default gems to be installed from
the .gem packages instead from the expanded sources.
Notes:
Merged: https://github.com/ruby/ruby/pull/2933
|
|
MiniTest::Unit (superclass of Test::Unit::Runner) does not has
default seed parameter, but assigned after initializing.
However some tests use MiniTest::Unit without setup of seed option
and it cases unexpected test failures. To solve this issue, add
default seed parameter 42.
|
|
To avoid `srand(0)` effect in the other tests, call `srand($seed)`
at the beginning of each test (setup).
[Feature #16655]
|
|
This reverts commit 7c1553e91db07fed4fed3287b50304f1c69a685d.
It breaks some tests.
|
|
To avoid `srand(0)` effect in the other tests, call `srand($seed)`
at the beginning of each test (setup).
[Feature #16655]
|
|
`--seed N` option is just ignored so respect this option.
[Feature #16655]
Also making "Run options" display pretty.
|
|
Fixed a TODO in b9007b6c548f91e88fd3f2ffa23de740431fa969
|
|
Simply use `File.basename` to remove the directory name (and
suffix), instead of `gsub` which can replace unintended parts.
|
|
It is not necessary to strip the `destdir` prefix every iteration, when
it can be done just once.
Notes:
Merged: https://github.com/ruby/ruby/pull/2515
|
|
.gemspec files specifies not just `bin`, but also other directories.
Notes:
Merged: https://github.com/ruby/ruby/pull/2515
|
|
The local `path` variable does not provide any additional value and was
kept around just for clarity for easier review of the `extrac_files`
method move.
Notes:
Merged: https://github.com/ruby/ruby/pull/2515
|
|
This just gets the `RbInstall::DirPackage` closer by functionality to
`Gem::Package`.
Notes:
Merged: https://github.com/ruby/ruby/pull/2515
|
|
rbinstall is using `$script_mode` and `$prog_mode`. However, the
`$script_mode` fallbacks to `$prog_mode` if not provided.
However, RubyGems do not distinguish between `$script_mode` and
`$prog_mode`:
https://github.com/rubygems/rubygems/blame/92892bbc3adba86a90756c385433835f6761b8da/lib/rubygems/installer.rb#L196
https://github.com/rubygems/rubygems/blame/92892bbc3adba86a90756c385433835f6761b8da/lib/rubygems/installer.rb#L515
https://github.com/rubygems/rubygems/blame/92892bbc3adba86a90756c385433835f6761b8da/lib/rubygems/installer.rb#L543
Comparing the usage of `$script_mode` and `$prog_mode`, it seems that
the `$script_mode` should be used where RubyGems expects `$prog_mode`.
Notes:
Merged: https://github.com/ruby/ruby/pull/2515
|
|
1. This is similar to what RubyGems does and it is less magic [[1]].
2. It avoids deprecated code paths in RubyGems [[2]].
[1]: https://github.com/rubygems/rubygems/blob/92892bbc3adba86a90756c385433835f6761b8da/lib/rubygems/installer.rb#L151
[2]: https://github.com/rubygems/rubygems/blob/92892bbc3adba86a90756c385433835f6761b8da/lib/rubygems/installer.rb#L187
Notes:
Merged: https://github.com/ruby/ruby/pull/2515
|
|
Extract bundled gems under ".bundle/gems" and get rid of
duplication which cause constant redefinition warnings at
`test-all` after `extract-gems` and `test-bundler`.
Notes:
Merged: https://github.com/ruby/ruby/pull/2922
|
|
|
|
rss 0.2.9 is tagged without the "v" prefix.
|
|
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.
Notes:
Merged: https://github.com/ruby/ruby/pull/2888
|