[#97319] [Ruby master Feature#16667] Allow parameters to Symbol#to_proc and Method#to_proc — jgomo3@...

Issue #16667 has been reported by jgomo3 (Jes=FAs G=F3mez).

10 messages 2020/03/01

[#97344] [Ruby master Feature#16670] Reverse order of `expression` in `pattern` for 1-line pattern matching while it's still experimental — ttilberg@...

Issue #16670 has been reported by ttilberg (Tim Tilberg).

9 messages 2020/03/03

[#97355] [Ruby master Misc#16671] BASERUBY version policy — ko1@...

Issue #16671 has been reported by ko1 (Koichi Sasada).

10 messages 2020/03/04

[#97359] [Ruby master Bug#16672] net/http leaves original content-length header intact after inflating response — justin.reid@...

Issue #16672 has been reported by jmreid (Justin Reid).

15 messages 2020/03/04

[#97390] [Ruby master Bug#16677] Negative integer powered (**) to a float number results in a complex — camille.drapier@...

Issue #16677 has been reported by CamilleDrapier (Camille Drapier).

25 messages 2020/03/07

[#97410] [Ruby master Bug#16680] [Breaking Change] Ruby 2.7 not support symlinks folder in $LOAD_PATH to work with autoload. — vil963@...

SXNzdWUgIzE2NjgwIGhhcyBiZWVuIHJlcG9ydGVkIGJ5IHp3OTYzIChXZWkgWmhlbmcpLg0KDQot

8 messages 2020/03/07

[#97416] [Ruby master Bug#16682] Ruby 2.7.0p0 crash on exit if there is an active RUBY_INTERNAL_EVENT_GC_EXIT tracepoint — jean.boussier@...

Issue #16682 has been reported by byroot (Jean Boussier).

16 messages 2020/03/09

[#97448] [Ruby master Feature#16688] Allow #to_path object as argument to system() [email protected]

Issue #16688 has been reported by Dan0042 (Daniel DeLorme).

12 messages 2020/03/11

[#97528] [Ruby master Misc#16693] DevelopersMeeting20200410Japan — mame@...

Issue #16693 has been reported by mame (Yusuke Endoh).

12 messages 2020/03/16

[#97536] [Ruby master Bug#16694] JIT vs hardened GCC with PCH — v.ondruch@...

Issue #16694 has been reported by vo.x (Vit Ondruch).

11 messages 2020/03/18

[#97538] [Ruby master Bug#16695] Stack consistency error when using the return value — s.wakeup31@...

Issue #16695 has been reported by s4ichi (takamasa saichi).

10 messages 2020/03/18

[#97554] [Ruby master Bug#16697] Hash.ruby2_keywords_hash?(value) should support any object — eregontp@...

Issue #16697 has been reported by Eregon (Benoit Daloze).

12 messages 2020/03/19

[#97588] [Ruby master Feature#16739] Allow Hash#keys and Hash#values to accept a block for filtering output — jacobevelyn@...

SXNzdWUgIzE2NzM5IGhhcyBiZWVuIHJlcG9ydGVkIGJ5IGphY29iZXZlbHluIChKYWNvYiBFdmVs

7 messages 2020/03/24

[#97609] [Ruby master Bug#16740] Deprecating and removing the broken Process.clock_getres — eregontp@...

Issue #16740 has been reported by Eregon (Benoit Daloze).

14 messages 2020/03/28

[#97621] [Ruby master Bug#16743] problem with multi threading [BUG] Segmentation fault — pauloo.jansen@...

Issue #16743 has been reported by paulorja (paulo jansen).

12 messages 2020/03/29

[#97629] [Ruby master Feature#16744] Flag to load current bundle without using bundle exec — headius@...

Issue #16744 has been reported by headius (Charles Nutter).

11 messages 2020/03/30

[ruby-core:97384] [Ruby master Bug#16676] `#hash` can change Hash object from ar_table to st_table

From: ko1@...
Date: 2020-03-06 19:08:58 UTC
List: ruby-core #97384
Issue #16676 has been updated by ko1 (Koichi Sasada).

Status changed from Open to Closed

4c019f5a626523e99e2827ed917802e3097c380d
c3584dfacce4d0f2058d8403de6fdce4fd4d686b

----------------------------------------
Bug #16676: `#hash` can change Hash object from ar_table to st_table
https://bugs.ruby-lang.org/issues/16676#change-84510

* Author: ko1 (Koichi Sasada)
* Status: Closed
* Priority: Normal
* Assignee: ko1 (Koichi Sasada)
* Target version: 2.8
* Backport: 2.5: UNKNOWN, 2.6: UNKNOWN, 2.7: UNKNOWN
----------------------------------------
# Points

* Hash representation `ar_table` has a problem from Ruby 2.6 (need to backport)

# Problem

From Ruby 2.6 `Hash` object is represented by `ar_table` (for small (<=8) pairs) and `st_table`.
To lookup hash table, `Object#hash` is called to get a hash value, and this method call can modify Hash object.
It means that the code assumes the Hash representation is a `ar_table`, but calling the `#hash` method, it can be a `st_table`.

I believe nobody modify Hash table from `#hash` method (it should be pure, no-side effect method), but multi-threads can modify a Hash object, it can be exposed.

I will commit a patch soon.

# Workaround

To avoid this issue, making a hash table in `st_table`, for example:

```
h = {} # ar_table
10.times{|i| h[i] = i} # st_table because the number of pair == 10 (> 8)
h.clear
```




-- 
https://bugs.ruby-lang.org/

Unsubscribe: <mailto:[email protected]?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>

In This Thread