Activity
From 10/10/2020 to 10/16/2020
10/16/2020
-
11:39 PM Feature #17176: GC.auto_compact / GC.auto_compact=(flag)
- I've updated the patches on the PR to reflect the new API. Now you can do `GC.auto_compact = true/false` and `GC.auto_compact`.
-
11:18 PM Revision f6661f50 (git): sync RClass::ext::iv_index_tbl
- iv_index_tbl manages instance variable indexes (ID -> index).
This data structure should be synchronized with other ractors
so introduce some VM locks.
This patch also introduced atomic ivar cache used by
set/getinlinecache instructions... -
- (1) nobody uses it (gem-codesearch)
(2) the data strucuture will be changed. -
-
03:28 PM Feature #17171: Why is the visibility of constants not affected by `private`?
- At the same time it would be worth reconsidering #16276, which would allow `private{ X = 42 }` right now without a deprecation warning or having to use `private_constant`
-
02:49 PM Feature #17171: Why is the visibility of constants not affected by `private`?
- If `private` affects constants, are constants private after `protected`?
```ruby
class Foo
protected
P1 = 1 # private constant or not?
private
protected
P2 = 2 # private constant or not?
end
``` -
05:26 AM Feature #17171: Why is the visibility of constants not affected by `private`?
- Deprecating on access (first time only) would be less noisy and sounds easier to implement indeed 👍
-
04:19 AM Feature #17171: Why is the visibility of constants not affected by `private`?
- marcandre (Marc-Andre Lafortune) wrote in #note-7:
> So can we envision issuing a warning for two release cycles?
> ...
I think if we do this, we should only warn on access (i.e. `Foo::X`), not definition. This would require internal ... -
- So can we envision issuing a warning for two release cycles?
Is it feasible to issue a warning for constants that are currently public that would be made private this way?
Difficult being that call to `private_constant` happens aft... -
03:22 PM Bug #17264 (Closed): BigDecimal exponentiation cannot be used with #** method
- As stated in Bug #17214 (https://bugs.ruby-lang.org/issues/17214) when exponentiating a BigDdecimal number even when using small numbers, a precision argument must be passed to the operation or the operation will return wrong result, 222...
- 03:08 PM Revision 26e8db6b (git): * 2020-10-17 [ci skip]
-
03:07 PM Revision ac803ab5 (git): test/rinda/test_rinda.rb: Add more debugging code
- in addition to de5e8d0e3bc3cc39487ffc9d9c15642b6881cd54
-
02:24 PM Revision 0d17cdd0 (git): Abort on system stack overflow during GC
- Buggy native extensions could have mark functions that cause stack
overflow. When a stack overflow happens during GC, Ruby used to recover
by raising an exception, which runs the interpreter. It's not safe to
run the interpreter during G... -
10:49 AM Bug #17220 (Third Party's Issue): Rails Active Job integration test fails with Ruby 3.0.0 since 2038cc6cab6ceeffef3ec3a765c70ae684f829ed
- `getaddrinfo_a` seems to create a thread, and it will be lost by `fork`.
I'm not sure if this is a glibc's spec or bug.
This code demonstrates the behavior without Ruby.
```C
#define _GNU_SOURCE
#include <netdb.h>
#include <std... -
- https://rubyci.org/logs/rubyci.s3.amazonaws.com/rhel8/ruby-master/log/20201016T063003Z.fail.html.gz
```
1) Error:
Rinda::TestRingServer#test_do_reply:
Timeout::Error: timeout
/home/chkbuild/chkbuild/tmp/build/20201016T063003Z/ruby/... -
06:19 AM Bug #17263: Fiber context switch degrades with number of fibers, limit on number of fibers
- I've been meaning to revisit the x64 implementation to make it more memory friendly.
I played around with it today in my spare time. Here is updated implementation:
```
.globl PREFIXED_SYMBOL(SYMBOL_PREFIX,coroutine_transfer)
PRE... -
- On my computer, I found the following.
I changed your script to run with the given number of fibers as an argument.
```
> sudo perf stat -e page-faults,cpu-cycles:u,cpu-cycles:k ./ruby ../test.rb 10000
fibers: 10000 rss: 143252 count: ... -
- * [Feature #17171] Have `private` affect constants and singleton methods definitions too (marcandre)
* Privacy is important for gems in particular. `private_constant` is verbose to use.
* The constants and methods affected would ha... -
- Looking at the `ENV.replace` implementation, the reason it doesn't call `clear` is it isn't thread-safe. `ENV.replace` tries to set all ENV keys in the replacement hash (the argument), then remove keys for ENV that aren't in the replace...
-
-
- Does this error also occur when building Ruby 2.7.2 in your environment?
10/15/2020
-
11:57 PM Bug #17220: Rails Active Job integration test fails with Ruby 3.0.0 since 2038cc6cab6ceeffef3ec3a765c70ae684f829ed
- Here's a small script that reproduces the problem:
```ruby
require "socket"
Socket.getaddrinfo("localhost", nil)
pid = fork do
Socket.getaddrinfo("localhost", nil)
end
Process.wait pid
```
In Rails' test suite, the `... - 11:37 PM Revision c34539d0 (git): * 2020-10-16 [ci skip]
-
11:37 PM Bug #16695 (Closed): Stack consistency error when using the return value
- Applied in changeset commit:git|ce7a053475cbebfb2f3e5ed6614e0ba631541917.
----------
Adjust sp for `x = false; y = (return until x unless x)` [Bug #16695] -
-
-
07:10 PM Feature #16986: Anonymous Struct literal
- I think a new syntax for this is way too heavy for such a small thing.
```ruby
s = Struct.new(:a, :b).new(0, 0)
```
is easy enough and already works.
And refactoring that to make it efficient for many instances is just giving it... -
- From the aspect of "writing-well", `${a: 0, b: 0}` was happy with me (but I understand some (many?) people doesn't like `$`), but `%struct{a: 0, b: 0}` is longer. `Struct(a:0, b:0)` is shorter and clearer...
-
- marcandre (Marc-Andre Lafortune) wrote in #note-55:
> Seems to me that `s = {a: 0, b: 0}` is a good solution here. If you mistype `aa`, you *want* an error, so I don't see the issue.
In this case, yes. So IDE support in future? (clea... -
- ko1 (Koichi Sasada) wrote in #note-54:
> Of course, sometimes I use `s = {a: 0, b: 0}`, but sometimes I mistype as `s[:aa] += val #=> error because of nil+val`
Seems to me that `s = {a: 0, b: 0}` is a good solution here. If you misty... -
- Named tuple example:
On my scripting sometimes I use an array as a tuple like that:
```ruby
s = [0, 0] # s is "statistics" and [A's count, B's count]
# I want to prohibit extension of `s` (s[3] = 0), but there is no way.
dat... -
- I can confirm I can reproduce your problem. As an experiment, I did:
```
first.transfer
count = 0
```
as warmup to see if it was some kind of lazy allocation issue, but it didn't seem to help. I'll have to take a look at the i... -
- Regarding "can't set a guard page" it's because of your system is limiting the number of memory mapped segments. Each fiber stack requires a guard page and this is considered a separate memory map entry.
Try increasing it:
```
sysctl -... -
10:04 AM Bug #17263 (Closed): Fiber context switch degrades with number of fibers, limit on number of fibers
- I'm working on developing [Polyphony](https://github.com/digital-fabric/polyphony), a Ruby gem for writing
highly-concurrent Ruby programs with fibers. In the course of my work I have
come up against two problems using Ruby fibers:
... -
08:19 AM Revision b68c22b3 (git): Partly reverted test failing with https://github.com/rubygems/rubygems/pull/3921
-
-
- Please backport following fixes for Encoding.
* commit:5f7be6243a5b support multi-run for test/ruby/test_encoding.rb
* commit:921916ff9e29 Isolate the test for Encoding#replicate
* commit:52cdf400efae Workaround of instance variable... -
- https://hackerone.com/reports/954433
-
10/14/2020
- 11:44 PM Revision f0c7a05b (git): * 2020-10-15 [ci skip]
-
- Update the lldb script so it can mostly recover a Ruby stack trace from
a core file. It's still missing line numbers and dealing with CFUNCs,
but you use it like this:
```
(lldb) rbbt ec
rb_control_frame_t TYPE
0x7f6fd6555fa0 EVAL ... -
- RubyVM.stat[:global_method_state] is no longer available so
this test doesn't check any more. -
- Now ruby_vm_global_method_state is not used so let's remove it.
-
- * [Feature #17259] Kernel#warn should ignore <internal: entries (eregon)
* Currently, `warn('message', uplevel: n)` can show `<internal:kernel>:90: warning: message` instead of `calling_file.rb:42: warning: message`.
* https://gith... -
- generic_ivtbl is a process global table to maintain instance variables
for non T_OBJECT/T_CLASS/... objects. So we need to protect them
for multi-Ractor exection.
Hint: we can make them Ractor local for unshareable objects, but
no... -
- (1) recorded_lock_rec > current_lock_rec should not be occurred
on rb_ec_vm_lock_rec_release().
(2) should be release VM lock at EXEC_TAG(), not POP_TAG().
(3) some refactoring. -
- ref bf3b2a43741e4f72be21bc6acf24d37e7fcff61c
-
-
-
- Applied in changeset commit:git|102c2ba65f1fa2a6cdbaaa7d2b466aabfc50e5d7.
----------
freeze Encoding objects
Encoding objects can be accessed in multi-ractors and there is no
state to mutate. So we can mark it as frozen and shareable.
... -
- Encoding objects can be accessed in multi-ractors and there is no
state to mutate. So we can mark it as frozen and shareable.
[Bug #17188] -
- enc_table which manages Encoding information. rb_encoding_list
also manages Encoding objects. Both are accessed/modified by ractors
simultaneously so that they should be synchronized.
For enc_table, this patch introduced GLOBAL_ENC_TABL... -
- If a ractor getting a VM lock (monitor) raises an exception,
unlock can be skipped. To release VM lock correctly on exception
(or other jumps with JUMP_TAG), EC_POP_TAG() releases VM lock.
10/13/2020
-
08:27 PM Bug #17215: Backport for arm64 optimizations that exist for power/x86
- Thank you for the explanation. I remember the backporting rule now. I am sorry for disturbing you.
-
- jaruga (Jun Aruga) wrote in #note-6:
> I am asking the backport to Ruby project's 2.7 branch.
> ...
Backport issues are closed after the appropriate backport marker has been set, similarly to how regular bug fixes are closed after the ... -
- I am asking the backport to Ruby project's 2.7 branch.
If you use the patch on Fedora, why does it connect for you to close this ticket's status?
-
04:11 PM Bug #17215 (Closed): Backport for arm64 optimizations that exist for power/x86
- I am going to use this patch on Fedora, if it helps:
https://src.fedoraproject.org/rpms/ruby/blob/master/f/ruby-3.0.0-preview1-Enable-arm64-optimizations-that-exist-for-power-x86.patch -
- PR: https://github.com/ruby/ruby/pull/3647
-
- Interesting :)
Having it on `Thread` sounds nice to me.
I wonder how limiting it is to restrict to only shareable values for TVar.
Is there actually any type usable for TVar except Integer?
A frozen String could only be swapped w... - 06:18 PM Revision d7de342e (git): * 2020-10-14 [ci skip]
-
-
04:44 PM Feature #17260: Promote pattern matching to official feature
- marcandre (Marc-Andre Lafortune) wrote in #note-8:
> We can document that this constraint is projected in the future.
+1 -
- palkan (Vladimir Dementyev) wrote in #note-7:
> I have a little concern regarding the potential changes to the current specification that we might need to introduce to improve the performance.
> ...
I'm glad to read this, I asked for [... -
- ktsj (Kazuki Tsujimoto) wrote:
> I propose to promote pattern matching to official feature.
> ...
I have a little concern regarding the potential changes to the current specification that we might need to introduce to improve the perfo... -
- 04:41 AM Revision 10e09cca (git): * 2020-10-13 [ci skip]
-
10/12/2020
-
- matz (Yukihiro Matsumoto) wrote in #note-1:
> I have a secret plan to make right hand assignment a replacement for single line pattern match.
I find myself liking this idea a lot.
In #13683 nobu suggested pattern matching as a rep... -
12:16 PM Feature #17260: Promote pattern matching to official feature
- That's true. Hmm.
Matz.
-
- It disallows assignments to other than local variables.
https://github.com/nobu/ruby/runs/1240626799#step:15:315
```
SyntaxError: (eval):1: syntax error, unexpected instance variable
def self.inc(x) = x + 1 => @x
... -
- Seems like so.
Matz.
-
- matz (Yukihiro Matsumoto) wrote in #note-1:
> I have a secret plan to make right hand assignment a replacement for single line pattern match. But it's another story.
Maybe, is this the kind of thing?
https://github.com/nobu/ruby/p... -
- Agreed.
I have a secret plan to make right hand assignment a replacement for single line pattern match. But it's another story.
Matz.
-
04:16 PM Bug #17221: Relax the Fiber#transfer's limitation
- It seems like both "a" and "the" work here. I might say, "cannot transfer to a yielding Fiber" or "attempted transfer to a yielding Fiber."
-
- Applied in changeset commit:git|bf3b2a43741e4f72be21bc6acf24d37e7fcff61c.
----------
relax Fiber#transfer's restriction
Using Fiber#transfer with Fiber#resume for a same Fiber is
limited (once Fiber#transfer is called for a fiber, the ... -
- BTW I already got Matz's approval so I'll merge it as soon as possible.
-
- Next English question :)
"attempt to transfer to a yielding fiber" or "attempt to transfer to the yielding fiber" (a/the) on error message (error message when calling `fib.transfer` and `fib` is yielding). -
- Thanks!
-
- on Android
-
- Using Fiber#transfer with Fiber#resume for a same Fiber is
limited (once Fiber#transfer is called for a fiber, the fiber
can not be resumed more). This restriction was introduced to
protect the resume/yield chain, but we realized that it... -
- Attempting to create a hard link raises EACCES
-
- See http://ruby-concurrency.github.io/concurrent-ruby/1.1.4/Concurrent/TVar.html
>A TVar is a transactional variable -
- What does TVar mean?
-
- ## Abstract
I propose Software transactional memory (STM) for threads and ractors.
Implementation is here: https://github.com/ruby/ruby/pull/3652
The interface is similar to concurrent-ruby, but not the same.
http://ruby-concur... - 06:16 AM Revision 6527411f (git): [ci skip] Minor documentation fix.
- Add missing period.
-
05:20 AM Revision e8d03c9a (git): change rb_ractor_queue to ring buffer
-
04:56 AM Revision c6652f22 (git): ractor.rb - indent comment code [ci skip]
-
- It uses random to determine if the bignum is sparse or not.
It is arguable if three-digit samples are enough or not to determine it,
but anyway, consuming Random source implicitly is not good.
I introduced the random sampling mechanism,... -
- Applied in changeset commit:git|8a39e6d6539bd37100cbbfb88916b853f444f771.
----------
bignum.c (rb_int_powm): Integer#pow(0, 1) should return 0
... instead of 1 because it requires "modulo 1". [Bug #17257] -
- I agree that it is a bug. The pseudocode of modular exponentiation in Wikipedia includes `if modulus = 1 then return 0` as the first check.
https://en.wikipedia.org/wiki/Modular_exponentiation#Pseudocode
I agree that there are few... -
- ... instead of 1 because it requires "modulo 1". [Bug #17257]
- 04:08 AM Revision 13366982 (git): Respectively, instead of respentively
10/11/2020
-
06:46 PM Bug #17257: Integer#pow(0, 1) returns 1, which is incorrect
- sawa (Tsuyoshi Sawada) wrote in #note-10:
> I cannot follow the discussion because of the expression "multiplying `x % m` to 1 `y` times." What does that mean?
"multiplying `x % m` to 1 `y` times"の意味は、「1に対して`(x % m)`を`y`回乗じること(又は、その値)」で... -
01:18 PM Bug #17257: Integer#pow(0, 1) returns 1, which is incorrect
- The point of undefinition is:
There is nothing definitely right or wrong.
Neither 0 nor 1 is wrong; it is just a different point of view.
You are thinking that 0 ** 0 == 1 is useful because you are looking from the perspective of x ... -
12:09 PM Bug #17257: Integer#pow(0, 1) returns 1, which is incorrect
- I cannot follow the discussion because of the expression "multiplying `x % m` to 1 `y` times." What does that mean?
-
- `x.pow(y, m)` doesn't equal multiplying `x % m` to 1 `y` times.
```ruby
p 3.pow(2, 4) #=> 1
p 1 * (3 % 4) * (3 % 4) #=> 9
```
It is the last to mod.
The result should be less than modulo.
Method `pow` should be simply defined as `(x**... -
- Wait a minute, I just changed my mind.
Actually since `0.pow(0)` is mathematically undefined.
See WolframAlpha for references:
https://www.wolframalpha.com/input/?i=0%5E0
`0.pow(0, 1)` should be undefined too, so do `1.pow(0, 1)... -
- So the proposal is to change the document from
"Returns (modular) exponentiation"
to
"multiplying x % m to 1 y times"?
IMHO `% m` is the space, `** y` is the operation performed inside the space, so the result should be within the ... -
- I agree that the behavior and documentation **may** be inconsistent, and think one option is to fix the latter.
- 04:53 PM Revision 20d78fdd (git): * 2020-10-12 [ci skip]
-
- Also unwrap NODE_RIPPER to check the method name.
-
12:06 PM Misc #17200: DevelopersMeeting20201026Japan
- * [Feature #17260] Promote pattern matching to official feature
* I'd like to remove single line pattern matching and experimental warnings.
-
- I propose to promote pattern matching to official feature.
The current specification is basically fine, but I'd like to reconsider single line pattern matching (`expr in pat`) and suggest removing it once in 3.0.
-
- nobu (Nobuyoshi Nakada) wrote in #note-3:
> Essentially `uplevel:` option is useless in some cases.
> ...
Yes, some libraries might want to emit warnings and ignore their own `my_gem/lib/`.
That however would need to make `warn` somehow ... -
- Essentially `uplevel:` option is useless in some cases.
For instance, it cannot work to skip frames in the same file or directory. -
06:42 AM Bug #17211: Test failures in ruby2.7.2 and ruby3.0~preview1
- ruby_2_7 d1ba5545513b68d39ca29b578a42bd8d48a7804e merged revision(s) 9718ff62c12c07ecf7f0e234343dca76ee1aa51d,20ad1017017ea736667d86fa0250dc1a39daefa1.
-
- Show stdout and stderr when history tests fail
Remove system method for E2E testing because depends on ruby command -
- I'm afraid that I did backport the changeset d23d2f3f6fbb5d787b0dd80675c489a692be23e2 solely at 828c34e58b63d64558ec0f2d1d7ae401c5e6b21f.
I applied the remaining part of the patch into ruby_2_7 at 48ac73769772317d6c3f864f087ef930a47120d9. -
-
-
-
03:13 AM Bug #17221: Relax the Fiber#transfer's limitation
- Eregon (Benoit Daloze) wrote in #note-11:
> I believe it's `a yielding Fiber` (the first sound in yield-ing is not a vowel sound)
A Yes indeed for this question. (not an yes :-) -
10/10/2020
-
- It's a bug.
Additional tests look good. -
- This is undefined
`0.pow(0, 1)`
https://www.wolframalpha.com/input/?i=0%5E0
This is clearly out of the space (thus a bug)
`1.pow(0, 1)`
https://www.wolframalpha.com/input/?i=%281%5E0%29+mod+1
It is practically pointless to mod ... -
- According to RDoC, `x.pow(y, m)` equals `(x**y) % m`.
In fact, `x.pow(y, m)` returns same value as `(x**y) % m` except for `x.pow(0, 1)`.
```ruby
p (12**0) % 1 #=> 0
p 12.pow(0, 1) #=> 1
``` -
- I'm not sure if it should be 0.
As `x.pow(y, m)` equals multiplying `x % m` to 1 `y` times, isn't 1 ok when `y == 0`? -
- I agree with removing the Rubygems override of `Kernel#warn` (the only Kernel method Rubygems should override is `require`). Skipping `<internal:` entries for `uplevel` makes sense to me.
-
- `Kernel#warn` currently does not skip `<internal:` entries from core library methods defined in Ruby.
This can cause rather unhelpful locations to be used for warnings.
For instance:
```
$ ruby -v --disable=gems -e 'def deprecated;... -
- https://github.com/ruby/io-console/commit/f84e6abcce
-
- When `min: 0` is given to `IO#getch` with `time:` option, it is
expected to return nil if timed out, and needed for source code
the compatibility with unixen platforms.
https://github.com/ruby/io-console/commit/a2afbe72bd -
- Fixed TypeError when IO#getch timed out
`rb_io_wait` returns a bit-flags Integer representing available
events, or Qfalse if timed out. Also the result of `NUM2INT` is
not a `VALUE`.
```
$ ./bin/ruby -v -rio/console -e "p IO.console.g... - 04:28 PM Revision e10d4dce (git): * 2020-10-11 [ci skip]
-
- * Fixed backward compatibility.
* Added missing `rb_scheduler_timeout` declaration.
https://github.com/ruby/io-console/commit/813806079f -
- * Keep Ractor#recv/Ractor.recv as an alias for now.
-
10:14 AM Revision 9eccf071 (git): Update RBS to 0.13.1 (#3645)
-
08:52 AM Revision 5e120a23 (git): Improve doc in rb_class_real doc (#3637)
- 06:32 AM Revision df250070 (git): Fixed typo in comment
- alway -> always