Activity
From 02/20/2020 to 02/26/2020
02/26/2020
-
11:23 PM Feature #16644: qualified const init (self::CONST1 = 1) should be allowed in methods
- a qualified const init is so similar to meta-programming `const_set` that there's no reason to block it
```ruby
module Mod1
end
def define_consts(mod)
# since this allowed
mod.const_set(:CONST1, :CONST1) unless mod.const... -
07:40 PM Bug #16654: Segfault in rb_str_hash
- The application I'm reproducing it on is here: https://github.com/codetriage/codetriage
I think you should be able to "bundle install" and then use my script above and it would reproduce.
-
07:11 AM Bug #16654: Segfault in rb_str_hash
- Nobu, This looks like Mac to me maybe llvm related?
-
05:29 AM Bug #16654 (Feedback): Segfault in rb_str_hash
- I couldn't reproduce it with a simple file.
Could you try with the master, and could your "config/environment.rb" be shared? -
07:06 PM Revision 6b30638b (git): MiniTest::Unit.options has default :seed
- 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 th... -
- To avoid `srand(0)` effect in the other tests, call `srand($seed)`
at the beginning of each test (setup).
[Feature #16655] -
- This reverts commit 7c1553e91db07fed4fed3287b50304f1c69a685d.
It breaks some tests. -
- Applied in changeset commit:git|116c6320956ce80efcb1db849a190ebd7f8c8ad3.
----------
don't ignore `--seed` option
`--seed N` option is just ignored so respect this option.
[Feature #16655]
Also making "Run options" display pretty. -
- # Points
* call `srand($seed)` before each test. `$seed` is given seed number by `--seed N`
# Background
Some tests use `rand()` to produce random number to generate random test examples.
Example:
```
def test_local_va... - 05:32 PM Revision 8296da88 (git): * 2020-02-27 [ci skip]
-
- To avoid `srand(0)` effect in the other tests, call `srand($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. -
02:38 PM Bug #16647: Redefining $/ causes irb to misbehave
- Eregon (Benoit Daloze) wrote in #note-1:
> Out of curiosity, why do you overwrite $/?
> ...
I was following along an example from the book "The Well Grounded Rubyist, 3rd edition", that was illustrating how IO objects iterate. -
08:37 AM Revision 83705c42 (git): Add missing free on cc_entries
- Looks like an oversight in b9007b6c548f91e88fd3f2ffa23de740431fa969 and
7ec23593746c8ccabd6c005cc34dde77d564c6c9. -
- just in case.
-
- Fixed a TODO in b9007b6c548f91e88fd3f2ffa23de740431fa969
-
- Add a little more details in 7ec23593746c8ccabd6c005cc34dde77d564c6c9
-
- shared by ko1
-
- Simply use `File.basename` to remove the directory name (and
suffix), instead of `gsub` which can replace unintended parts. -
- It was unnecessary in b9007b6c548f91e88fd3f2ffa23de740431fa969
-
07:01 AM Bug #16521 (Closed): ST_DEBUG not working
- Applied in changeset commit:git|fbd7f08e922cdf1a4be7a71aa056659971c81dab.
----------
kill ST_DEBUG [Bug #16521]
This compile-time option has been broken for years (at least since
commit 4663c224fa6c925ce54af32fd1c1cbac9508f5ec, accordi... -
- This compile-time option has been broken for years (at least since
commit 4663c224fa6c925ce54af32fd1c1cbac9508f5ec, according to git
bisect). Let's delete codes that no longer work. -
- This compile-time option has been broken for years (at least since
commit 49369ef17316cd5d5819b038f286e1d951972b52, according to git
bisect). Let's delete codes that no longer works. -
- ISO/IEC 9899:1999 section 6.10.3 paragraph 11 explicitly states that
"If there are sequences of preprocessing tokens within the list of
arguments that would otherwise act as preprocessing directives, the
behavior is undefined."
rb_str_n... -
05:48 AM Feature #16652 (Assigned): Use RubyGems facilities to install Gems during Ruby installation
- 02:04 AM Revision 9d6d5315 (git): Cache destination dir.
- It is not necessary to strip the `destdir` prefix every iteration, when
it can be done just once. - 02:04 AM Revision b8a8fdba (git): Properly detect mode for binaries.
- .gemspec files specifies not just `bin`, but also other directories.
- 02:04 AM Revision e087b029 (git): Use class variable `@src_dir` instead of local `path`.
- 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. - 02:04 AM Revision c9fab1ac (git): Use local `{dir,prog,data}_mode` variables instead of globals.
- This just gets the `RbInstall::DirPackage` closer by functionality to
`Gem::Package`. - 02:04 AM Revision 6609940a (git): Use `$script_mode` instead of `$prog_mode` for gem inistallation.
- rbinstall is using `$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/rubyge... - 02:04 AM Revision e960ef6f (git): Use `Gem::Package` like object instead of monkey patching.
- 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... -
- Mingw has meant "mingw on cygwin", and msys has not been a supported platform.
-
- Fix up 66d1900423e6fb9774c2fe72dba8c2968b54d7ab, `RubyVM::MJIT` is
available since ruby 2.6. -
01:11 AM Bug #16632: Remove verbose warning on treating keyword splat as positional argument in Ruby 2.6
- Looking through the repository history, the patch attached to this issue is mostly implemented by commit:6424d316b993ecceb6f583ae476096274e304788, so that commit could be backported instead, though I'm not sure it applies without conflicts.
02/25/2020
-
- Does not reproduce on 2.5.5, and reproduces on 2.6.0
Unfortunately, doesn't reproduce with a simple script, e.g.:
```ruby
require "memory_profiler"
report = MemoryProfiler.report do
1_000_000.times { "allocate a string" }
end
report... -
- This reproduces on 2.7.0 as well (though the thing calling `rb_str_hash` was different, this time `objspace_each_objects`)
```
-- C level backtrace information -------------------------------------------
/Users/nateberkopec/.rubies/ruby... -
- Seen in ruby 2.6.5.
```
-- C level backtrace information -------------------------------------------
/Users/nateberkopec/.rubies/ruby-2.6.5/bin/ruby(rb_vm_bugreport+0x82) [0x10632d9c2]
/Users/nateberkopec/.rubies/ruby-2.6.5/bin/rub... -
06:54 PM Feature #16511: Staged warnings and better compatibility for keyword arguments in 2.7.1
- @Eregon thanks for your reply.
I won't bother making a point-by-point counterargument because it would be redundant and boring at this point. Suffice it to say that I find almost every single line you wrote to be either flawed or factua... - 06:43 PM Revision 672213ef (git): Document that Array#index and find_index are aliases [ci skip]
- 05:27 PM Revision 6c66761c (git): * 2020-02-26 [ci skip]
-
- To prevent CI failures like
http://ci.rvm.jp/results/trunk-mjit-wait@silicon-docker/2739995 -
04:13 PM Feature #16597: missing poll()
- Attaching test and fix.
This fixes the problem with not raising exception but now my patched interpreter tends to crash. Need to investigate. -
- Looks like the issue is using multiple domain names instead of IP addresses for the :nameserver option. Single domain name works, multiple IP addresses works, IP address and domain name works, multiple domain names does not work:
```... -
09:08 AM Bug #16653 (Rejected): Weird behaviour of Resolv module
- I have to deal with DNS request. Here is some examples.
When I ask A records for example.com at some public DNS servers I get the results:
`> Resolv::DNS.new(nameserver: ['8.8.8.8', '1.1.1.1']).getresources('example.com', Resolv::DNS... -
- 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`. -
10:34 AM Feature #15921: R-assign (rightward-assignment) operator
- I think `=>` is okay, but in case we want to use a keyword (ordinary word) for this feature, I think `as` would be good. `as` in SQL is similar to rightward assignment.
```ruby
(1..).lazy.map {|x| x*2} as x
p x.first(10)
``` -
-
- We would like to skip this extension library when libstdc++ is missing.
To avoid such situation let's use try_link instead of try_compile. -
- GC can invoke just after allocation of jit_unit->cc_entries so
it should be zero-cleared. -
04:30 AM Bug #12159 (Closed): Thread::Backtrace::Location#path returns absolute path for files loaded by require_relative
- I talked about this issue with nobu and ko1.
> Is it intentional? or a bug?
As @usa said, it is intentional according to nobu and ko1. Please use `#absolute_path`.
> ...
It may return relative path, too. IMO, "file name" inclu... -
- This code can generate CC objects so we only need to count
existing String objects. -
- ALLOC_N() can causes GC. Sometimes `mjit_copy_job_handler()`
can be called by mjit_worker thread which is not a Ruby thread,
so we need to prevent GC in this function. This patch has some
issues, but I introduce it to pass the tests. - 03:53 AM Revision 82d27604 (git): * 2020-02-25 [ci skip]
-
03:53 AM Revision fa1ec604 (git): Fix wrong documentation for return value of Pathname#fnmatch
-
- I agree with your proposal. It's difficult to handle for us continuously.
@matz
Can we remove it? -
02:24 AM Feature #16652: Use RubyGems facilities to install Gems during Ruby installation
- I glanced at rbinstall.rb briefly; for those interested, the link should be here:
https://github.com/ruby/ruby/blob/master/tool/rbinstall.rb
That seems quite old code, lots of global variables in it as well. Props to
everyone able...
02/24/2020
-
11:38 PM Feature #16648: improve GC performance by 5% with builtin_prefetch
- I ran the patch on some included GC benchmarks in the repo and it doesn't seem to be a pure win (built-ruby is the patched version):
```
$ make benchmark ITEM=gc_ COMPARE_RUBY=/opt/rubies/2.8.0-clean/bin/ruby OPTS=-v
/opt/rubies/2.6.... -
- There is a bug with handling POLLNVAL in the new code.
-
- This was already fixed in master at commit:f8a8f055123bc81fc13fa295b936504196df0da4, which changed it so ruby2_keywords does not pass empty keyword splats as hashes. Passing empty keyword splats as hashes is not needed in Ruby 3, it is ...
-
05:10 PM Feature #16652 (Closed): Use RubyGems facilities to install Gems during Ruby installation
- Currently, `tools/rbinstall` uses quite some custom code to install gems instead of using functionality available in RubyGems. I don't think this is good, because the code is non transparent ATM and it is duplicating functionality, which...
-
02:58 PM Bug #16647 (Closed): Redefining $/ causes irb to misbehave
- It was a problem with Reline. I fixed it by https://github.com/ruby/reline/pull/129.
> My personal opinion is we should deprecate all these variables affecting default separators,
> ...
I think so, but...maybe, so many beginners use ... -
09:03 AM Bug #16647: Redefining $/ causes irb to misbehave
- Related to https://bugs.ruby-lang.org/issues/14240
changing these variables is deprecated now, and it shows a warning -
06:55 AM Bug #16651 (Closed): Extensions Do Not Compile on Mingw64 with mingw32-make
- When mkmf.rb creates a Makefile for an extension, it will generate something that looks like this:
```makefile
srcdir = .
topdir = C/MSYS64/USR/LOCAL/ruby-2.7.0/include/ruby-2.7.0
hdrdir = $(topdir)
arch_hdrdir = C:/MSYS64/USR/LOC... -
- Using the latest version of mingw64 and msys2, neither ruby 2.6.5 nor 2.7.0 compile.
The issue is with miniruby.exe - it gets built but does not work. For example:
$ ./miniruby.exe -v
Returns nothing (I assume it is crashing but... -
- Applied in changeset commit:git|8b6e2685a4b7c7de905f7ed60c02ef85e0724754.
----------
Fixed symbol misused as ID
`rb_funcallv_public` and `rb_respond_to` require an `ID`, not a
`Symbol`. [Bug #16649] -
- `String#encode` document says fallback can be a Hash, a Proc, a Method or an object which has `[]` method.
> Sets the replacement string by the given object for undefined character. The object should be a Hash, a Proc, a Method, or a... -
- `rb_funcallv_public` and `rb_respond_to` require an `ID`, not a
`Symbol`. [Bug #16649] -
-
- rss 0.2.9 is tagged without the "v" prefix.
-
02/23/2020
-
11:53 PM Feature #16456: Ruby 2.7 argument delegation (...) should be its own kind of parameter in Method#parameters
- I'd definitely like to see this as well. It'd be useful for Sorbet since it uses the parameters method to reconstruct methods when generating scaffolds for gem methods and other Ruby code.
https://github.com/sorbet/sorbet/blob/29a967a... - 03:28 PM Revision f88d209b (git): * 2020-02-24 [ci skip]
-
03:28 PM Revision 281b3500 (git): Add pattern matching documentation
- Add separate doc/syntax/pattern_matching.rdoc, add
link to control_expressions.rdoc.
The documentation is "reverse-engineered" from Ruby 2.7
behavior and early preview presentations, and corrected
by pattern-matching feature author @k-tsj. -
02:24 PM Feature #16511: Staged warnings and better compatibility for keyword arguments in 2.7.1
- Thanks for writing that out.
I think in general tracking whether a Hash is keywords or positional dynamically makes everything complicated.
Long-term, I think it's clear we want to know syntactically whether a Hash is keyword argumen... -
- Out of curiosity, why do you overwrite $/?
My personal opinion is we should deprecate all these variables affecting default separators,
they tend to break any non-trivial program, not just IRB, and explicit seems always clearer. -
- Applied in changeset commit:git|6298ec2875a6f1a1e75698c96ceac94362f20bcf.
----------
Warn non-nil `$\` [Feature #14240] -
-
-
-
- This reverts commit c7b71af9e21b3f0b8fe413a3c20cf4bd633c40d0, as
an example in bundler expects untouch PATH.
02/22/2020
-
- These are two different things. Ruby always warns for assigning to an existing constant, regardless of using `const_set` or `CONST = ...`. It does not support using `CONST = ...` inside a method, though `const_set` is allowed. Similar...
-
- Can you please run the failing code from the command line and include the output? The awd files provided are binary files and I have no idea how to read them, and the diag file doesn't include the necessary information.
It appears th... -
06:03 PM Feature #16648 (Open): improve GC performance by 5% with builtin_prefetch
- The mark phase of non-incremental major GC is (I believe) dominated by pointer chasing. One way we can improve that is by prefetching cachelines from memory before they are accessed, to reduce stalls. I did some experimenting, and the ...
- 03:38 PM Revision f85ca4c8 (git): * 2020-02-23 [ci skip]
-
03:32 PM Revision 77dcc2c8 (git): hash.c: [DOC] fix examples for ENV.merge!
-
-
02:05 AM Bug #16646: OpenSSL::VERSION is not defined (moved to ruby/openssl)
- See https://github.com/ruby/openssl/issues/347
Please close. OpenSSL should have a new release soon... -
-
- Thanks for the fix nobu! (https://github.com/ruby/ruby/commit/5b29ea0845c14092abd866ce0183c52635bade4c)
-
- Applied in changeset commit:git|5b29ea0845c14092abd866ce0183c52635bade4c.
----------
Proc from Symbol needs a receiver
So its arity should be -2 instead of -1.
[Bug #16640]
https://bugs.ruby-lang.org/issues/16640#change-84337 -
-
-
- iseq->body->jit_unit is not available if USE_MJIT==0 .
-
- > treating non-Symbol keys as keywords if the method accepts arbitrary keywords is expected behavior
Yes, I was aware of that, it's just that I always thought of this as simply an _additional_ behavior; previously `**hash` would raise a... -
- Yes, treating non-Symbol keys as keywords if the method accepts arbitrary keywords is expected behavior, as mentioned in the Ruby 2.7.0 release notes (https://www.ruby-lang.org/en/news/2019/12/25/ruby-2-7-0-released/). This is a deliber...
-
- So its arity should be -2 instead of -1.
[Bug #16640]
https://bugs.ruby-lang.org/issues/16640#change-84337 -
- Unused CI (introduced from peephole optimization, etc) can be NULL
so introduce NULL check. -
01:25 AM Revision af12e386 (git): More ENV rdoc [ci skip]
-
- 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. Addi... -
- Applied in changeset commit:git|b9007b6c548f91e88fd3f2ffa23de740431fa969.
----------
Introduce disposable call-cache.
This patch contains several ideas:
(1) Disposable inline method cache (IMC) for race-free inline method cache
* ... - 12:59 AM Revision d7984d0f (git): * remove trailing spaces. [ci skip]
-
- 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 allo... -
- 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 w...
02/21/2020
-
-
- I can not answer your question but I believe matz mentioned the change several times in different
talks before. So a deprecation phase makes no real sense, unless matz and the core team wants
to postpone the change past ruby 3.0 - but ... -
- In ruby 2.7 non-symbol keys are now accepted as keywords, resulting in the following change in behavior:
```ruby
def foo(*a,**h) p [a,h] end
foo("a"=>42)
# [[{"a"=>42}], {}] in 2.6
# [[], {"a"=>42}] in 2.7
```
Given that the... -
- If I start irb and then do:
``` ruby
$/ = "asdf"
```
then every character I type shows a newline on the irb console. Previous versions of ruby do not have this behavior, and it is annoying. -
- There are several ways to fix this, maybe add to openssl.rb?
Noticed in several platforms... - 03:45 PM Revision 5a89c057 (git): * 2020-02-22 [ci skip]
-
- With refinements, too.
-
-
-
- https://github.com/ruby/irb/commit/dbc7b059c7
-
-
-
-
- test/net/http/test_https.rb: rename fixture methods to read_fixture
because it conflicts with test-unit gem. -
11:41 AM Bug #16337: kernel_gem.rb:68 - ThreadError
- I guess this is because you're not using the latest rubygems. Could you run `gem env version` and check that you have a rubygems version older than 3.1?
You need to request upgrading rubygems to google-fluentd, maybe here: https://git... -
- jeremyevans0 (Jeremy Evans) wrote in #note-3:
> Dan0042 (Daniel DeLorme) wrote in #note-2:
> ...
I personally think that it should not be equivalent. `obj.dig(a, b, c)` was introduced as a safe-navigation like `obj[a][b][c]`, so I thi... -
- Dan0042 (Daniel DeLorme) wrote in #note-2:
> Are you seriously telling me that you consider it normal and correct that `obj.dig(**kw)` is _not_ equivalent to `[obj].dig(0, **kw)` ???
It should be equivalent, if `dig` is defined corre... -
- Are you seriously telling me that you consider it normal and correct that `obj.dig(**kw)` is _not_ equivalent to `[obj].dig(0, **kw)` ???
I will note that your first example produces `2` whether keywords are passed with RB_PASS_CALLED... -
-
-
- As mentioned in https://bugs.ruby-lang.org/issues/16511#note-17
it seems very unlikely this triggers in practice, because indeed who would bother with an extra `**` when the method doesn't take keyword arguments?
Maybe for the case the... -
- The arity should actually be `-2` which means 1 required arguments + any number of optional arguments.
- 04:31 AM Revision 4643bf5d (git): * 2020-02-21 [ci skip]
-
02/20/2020
-
- This behavior is deliberate. `dig` should only be overridden if you want to be able to access objects stored inside the object. dig should accept `*args`, using the first element to look inside the object, and passing the remaining arg...
-
- The following behavior for `dig` looks very weird to me:
```ruby
o=Object.new
def o.dig(**kw) p [kw] end
o.dig(x:1)
#no warning
#[{:x=>1}]
[o].dig(0, x:1)
#warning: Using the last argument as keyword parameters is deprecat... -
- steveb3210 (Stephen Blackstone) wrote in #note-16:
> Unforunately Miller-Rabin is not a deterministic test for arbitrarily large n - its only the work in the paper https://arxiv.org/pdf/1509.00864.pdf that allows us to provide functional... -
07:19 PM Feature #16468: Switch to Miller-Rabin for Prime.prime?
- Attached is the latest diff.
-
- Dan0042 (Daniel DeLorme) wrote in #note-15:
> I think it would be interesting to expose the algorithm for larger numbers. So you could have `miller_rabin` which allows any integer, and `prime?` which checks the value and either call `mi... -
- I think it would be interesting to expose the algorithm for larger numbers. So you could have `miller_rabin` which allows any integer, and `prime?` which checks the value and either call `miller_rabin` or raise ArgumentError.
-
- - Add bounds check
- Add test
-
- ```rb
module Mod1
def self.define_consts
const_set(:CONST1, :CONST1)
# this is actual const re-assignment but only a warning
const_set(:CONST1, :CONST1)
# this is const initialization but becomes an ... -
- An extension that makes poll() features available and does not require integration into ruby core is eventmachine.
-
- to allow Intel C++ compiler to read ruby.h.
This is similar to 9930481a239fa7182429f3c3942ea033fb9b0320 -
-
-
-
- > my thoughts were: if respond_to_missing? returns true once, does this have any effect on this cache? even if it might later return false?
Just now, no effect. -
- ko1 (Koichi Sasada) wrote in #note-14:
> Hanmac (Hans Mackowiak) wrote in #note-13:
> ...
my thoughts were: if `respond_to_missing?` returns true once, does this have any effect on this cache? even if it might later return false? -
- I'll merge this patch tomorrow.
If you have any suggestion, please tell me.
-
- It has been deprecated since 2011.
https://github.com/ruby/rdoc/commit/5c2aa0f77d -
08:08 AM Revision c3cd9258 (git): Use `brew upgrade` instead of `brew update`
-
- 2.7 backport is done at commit:eaee631d5eda03f77ec39ec6fc1826a06ab06205.
- 07:36 AM Revision 0149863d (git): * 2020-02-20 [ci skip]
-
- when the next insn is already compiled by former branches.
-
- Get rid of double writing.
-
- Comma at the end of enum is allowed in C since C99. We can use them
internally. However when it comes to extension libraries, they could be
written in different C++ versions. We cannot assume sometihng. Public
headers shall keep compa... -
- Recent (since 2012 maybe?) MSVC ships a header named xkeycheck.h, which
(kindly!) aborts compilation on redefinition of C++ keywords. Let's not
define this in case of C++. -
- The `cxx.try_compile` command in this file kicks `cxx.have_devel?`
internally, which recursively calls `cxx.try_link` with a different
source code. We don't want that happen (the source code compiled in
this file must be the first one).... -
- Namely glibc has this macro on -DFORTIFY_SOURCE. We have to prevent
macro redefinition with different macro body. -
- `make leaked-globals` points out that this function is leaked. This has
not been detected in our CI because it is defined only when
VM_CHECK_MODE is nonzero.
Just make it static and everytihng goes well. -
- This is a ruby method with arity zero. Which means, this function takes
one argument (that is self). -
- GC_DEBUG=1 makes this test fail because it changes the size of struct
RVALUE. I don't think the test is useful then. Let's just skip. -
- Starting clang 11, casts between pointer and (narrower-than-pointer) int
are now warned. However all such thing in our repository are guaranteed
safe. Let's suppress the warnings. -
- Actually, looks like I didn't read the bug report closely enough. This is a different issue, and suggests that we should either recognize the frozen hash in the `ruby2_keywords` case and dup it, or remove the optimization.
The issue ... -
- I agree it is a bug. I'm not sure it is worth fixing. Basically, the reason behind it is the parser doesn't separate hash compilation from keyword argument compilation, and the optimization for `**{}` as the only keyword argument ends ...