[ruby-core:88009] [Ruby trunk Bug#14920] Backport r63857 to fix performance problem on Time class in MacOs systems

Issue #14920 has been updated by rafaelfranca (Rafael Fran巽a).


Thank you for changing the status and the backport information.

Yes, that patch seems safer, but when I applied only r63994 on top of v2_4_4 tag I still see the same results running the following script.

```
require 'benchmark'

def create_utc_time
  100_000.times { Time.utc(2018) }
end

p "Main Process: #{Benchmark.measure { create_utc_time } }"

pid = Process.fork do
  p "Fork Process: #{Benchmark.measure { create_utc_time }}"
end

Process.wait(pid)
``` 

```
$ ./build/bin/ruby utc_benchmark.rb
"Main Process:   0.100000   0.000000   0.100000 (  0.101757)\n"
"Fork Process:   9.810000  27.470000  37.280000 ( 37.981360)\n"
```

With only r63857 I get better results.

```
$ ./build/bin/ruby utc_benchmark.rb
"Main Process:   0.050000   0.000000   0.050000 (  0.052233)\n"
"Fork Process:   0.050000   0.000000   0.050000 (  0.064096)\n"
```

----------------------------------------
Bug #14920: Backport r63857 to fix performance problem on Time class in MacOs systems
https://bugs.ruby-lang.org/issues/14920#change-73028

* Author: rafaelfranca (Rafael Fran巽a)
* Status: Closed
* Priority: Normal
* Assignee: 
* Target version: 
* ruby -v: 
* Backport: 2.3: REQUIRED, 2.4: REQUIRED, 2.5: REQUIRED
----------------------------------------
Related to https://bugs.ruby-lang.org/issues/14435.

Both 2.4 and 2.5 are affected by this issue. Naruse-san commit fix it by not calling tzset() more than once when we know the year doesn't have leap seconds. 





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

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