[PHP-DEV] Re:

Mailing Lists php.internals
1

Hi Yuya, the “genius”,

Look—I don’t just talk.

First, when I introduced the RFC, before the discussion started, many people asked: “How is this different from other functions like array_search?” I provided the analysis, and there were many questions like that.

You told me to run a real-world test, and you saw that my function performed better than those functions.

All the code was designed based on proper principles, and I am sure you saw that yourself. You didn’t even leave a single comment during the discussion period because you knew you would face a completely technical response.

Just for your information: someone praised my work who holds a high position in php-src and whom I truly respect—meaning your opinion doesn’t really matter to me.

Respectfully,
Sepehr

در تاریخ چهارشنبه ۹ سپتامبر ۲۰۲۶، ۱۹:۴۷ youkidearitai <youkidearitai@gmail.com> نوشت:

No, If you are a skilled engineer, you should accept criticism.
Your skill is not enough php-src development.

To begin with, To unable resolving issue, You must compile and test on your PC.
Otherwise, Your problem is CI is red, but your test code is broken.
Not CI’s problem.
Fix below..

========DIFF========
001- === BENCHMARK RESULTS (%d iterations) ===
002- [Match at start]
003- C (array_str_contains) : %f ms
004- Userland (foreach) : %f ms
005- Speedup : %f x faster
001+ === BENCHMARK RESULTS (1,000 iterations)label

007- [Match in middle]
008- C (array_str_contains) : %f ms
009- Userland (foreach) : %f ms
010- Speedup : %f x faster
003+ [Match at start]
004+ C (array_str_contains) : 11.40 ms
005+ Userland (foreach) : 2.74 ms
006+ Speedup : 0.24x faster

012- [Match at end]
013- C (array_str_contains) : %f ms
014- Userland (foreach) : %f ms
015- Speedup : %f x faster
008+ [Match in middle]
009+ C (array_str_contains) : 12.85 ms
010+ Userland (foreach) : 5.94 ms
011+ Speedup : 0.46x faster

017- [No match (all)]
018- C (array_str_contains) : %f ms
019- Userland (foreach) : %f ms
020- Speedup : %f x faster
013+ [Match at end]
014+ C (array_str_contains) : 11.58 ms
015+ Userland (foreach) : 7.16 ms
016+ Speedup : 0.62x faster
017+
018+ [No match (all)]
019+ C (array_str_contains) : 6.47 ms
020+ Userland (foreach) : 8.22 ms
021+ Speedup : 1.27x faster
========DONE========
FAIL Benchmark array_str_contains vs userland implementations
[ext/standard/tests/array_str_contains_benchmark.phpt]

Furthermore, seems don’t increase speed. This approach is maybe wrong.

Anyway, You seems don’t hear from people.
Please hear from people.

Yuya

‪2026年9月9日(水) 20:56 ‫سپهر محمودی‬‎ <sepehrphpr@gmail.com>:‬

در تاریخ چهارشنبه ۹ سپتامبر ۲۰۲۶، ۱۳:۵۹ youkidearitai <youkidearitai@gmail.com> نوشت:

No,

‪2026年9月9日(水) 0:40 ‫سپهر محمودی‬‎ <sepehrphpr@gmail.com>:‬

Hi Yua,

Thanks for your follow-up and for the time you’ve dedicated to this.

As you know, we’ve been encountering Segmentation Fault or Timeout errors in the CI tests. I’ve reduced the benchmark iterations from 100,000 to 1,000 to see if the issue was due to excessive system load, but unfortunately, the CI error persists, and it seems the problem lies elsewhere.

The exact cause of this error is still unclear at this moment.

Following this function, I plan to work on graphics-related functions such as mask, str_mask, and others.

در تاریخ سه‌شنبه ۸ سپتامبر ۲۰۲۶، ۰۸:۰۹ youkidearitai <youkidearitai@gmail.com> نوشت:

Hi Sepehr

That PR fails your test.
Please compile and test from your PC before push the CI.
You can check php-src’s README
https://github.com/php/php-src/blob/master/README.md, If you don’t
understand how to build and test.


========DIFF========
001- === BENCHMARK RESULTS (%d iterations) ===
002- \[Match at start\]
003- C (array_str_contains) : %f ms
004- Userland (foreach) : %f ms
005- Speedup : %f x faster
001+ === BENCHMARK RESULTS (100,000 iterations) ===

007- \[Match in middle\]
008- C (array_str_contains) : %f ms
009- Userland (foreach) : %f ms
010- Speedup : %f x faster
003+ [Match at start]
004+ C (array_str_contains) : 577.66 ms
005+ Userland (foreach) : 13.74 ms
006+ Speedup : 0.02x faster

012- \[Match at end\]
013- C (array_str_contains) : %f ms
014- Userland (foreach) : %f ms
015- Speedup : %f x faster
008+ [Match in middle]
009+ C (array_str_contains) : 632.95 ms
010+ Userland (foreach) : 251.77 ms
011+ Speedup : 0.40x faster

017- \[No match (all)\]
018- C (array_str_contains) : %f ms
019- Userland (foreach) : %f ms
020- Speedup : %f x faster
013+ [Match at end]
014+ C (array_str_contains) : 625.46 ms
015+ Userland (foreach) : 482.63 ms
016+ Speedup : 0.77x faster
017+
018+ [No match (all)]
019+ C (array_str_contains) : 465.12 ms
020+ Userland (foreach) : 474.41 ms
021+ Speedup : 1.02x faster

Regards
Yuya

‪2026年9月8日(火) 2:21 ‫سپهر محمودی‬‎ <sepehrphpr@gmail.com>:‬

Hi Yuya,

Could you please take a look at the failed CI checks in my PR when
you have time?

https://github.com/php/php-src/pull/23512/checks

I’m getting 503 Service Unavailable errors when trying to access
the logs, so I haven’t been able to determine whether the failures
are related to my changes.

I’m particularly concerned about the Tracing JIT and FreeBSD failures.
I’d appreciate your help understanding them, or rerunning the failed
jobs if appropriate.

Thank you for your time.

Best regards,
Sepehr

–

Yuya Hamada (tekimen)

As I mentioned on GitHub,
https://github.com/php/php-src/pull/23512#issuecomment-5594220142
CI red is your test code is wrong, It’s not CI timeout.
You can confirm when build & test on your local PC.

I’m telling you straight, If you cannot hear from PHP internals
people, You are not enough ability on php-src yet.
You can choose other way to contribute in PHP.

Regards
Yuya

–

Yuya Hamada (tekimen)

Hello Yuya,
Mr. Hamada—several prominent figures in the PHP community have praised my work.
Please bear in mind that I am just preparing my very first function—and yes, it had a minor issue that I shared with you.
There was a time when you, too, were creating your first function and surely encountered a small glitch—unless, of course, you happen to be Larry Page. So, instead of being so critical, please offer your support and encouragement to someone who—arguably—is among the first one or two Iranians to reach this level through exceptional programming skills.
Best regards,
Sepehr

–

Yuya Hamada (tekimen)