!jngDrdMgndWibPCYsR:nixos.org

Nix PHP

72 Members
A room for PHP developers running on Nix21 Servers

Load older messages


SenderMessageTime
5 Feb 2023
@drupol:matrix.orgPolCan we have a look at : https://github.com/NixOS/nixpkgs/pull/21403419:20:25
6 Feb 2023
@drupol:matrix.orgPol Thanks etu ^^ The backport is ready as well: https://github.com/NixOS/nixpkgs/pull/214885 10:06:45
@drupol:matrix.orgPol ma27: These failures are unpredictable and related to the tests. I can't personally do anything about them. 10:07:11
@ma27:nicht-so.sexyma27Can't we skip them (on Darwin at least)?10:07:31
@drupol:matrix.orgPol I don't know, maybe we should ask etu or Jan Tojnar about that. 10:08:42
@etu:failar.nu@etu:failar.nuI guess we could yes10:34:50
@tgerbet:matrix.orgtgerbet

I checked the logs to compare with the previous bumps (https://github.com/NixOS/nixpkgs/pull/208928) and it looks like it is mostly the same tests that fail:

Blacklist (with glob, quote and comments) [tests/blacklist.phpt]
Bug #66338 (Optimization binding of class constants is not safely opcacheable) [tests/bug66338.phpt]
Bug #78106: Path resolution fails if opcache disabled during request [tests/bug78106.phpt]
ISSUE #115 (path issue when using phar) [tests/issue0115.phpt]
ISSUE #149 (Phar mount points not working this OPcache enabled) [tests/issue0149.phpt]
revalidate_path 01: OPCache must cache only resolved real paths when revalidate_path is set [tests/revalidate_path_01.phpt]
10:38:05
@tgerbet:matrix.orgtgerbet tests/blacklist.phpt is new if we compare the 2 PRs 10:38:40
@tgerbet:matrix.orgtgerbet * tests/blacklist.phpt failure is new if we compare the 2 PRs 10:38:50
@tgerbet:matrix.orgtgerbet

Excluding tests/blacklist.phpt all the tests that are failing are the ones starting a small CLI server

https://github.com/php/php-src/blob/php-8.2.2/ext/opcache/tests/bug66338.phpt#L33
https://github.com/php/php-src/blob/php-8.2.2/ext/opcache/tests/php_cli_server.inc

🤔

10:57:33
@tgerbet:matrix.orgtgerbet

__darwinAllowLocalNetworking for opcache is set to true but I'm not sure if the value is correctly set in when we build the extension

https://github.com/NixOS/nixpkgs/blob/8faf826ce5d79eeb731d77927cd850e53d7773d7/pkgs/top-level/php-packages.nix#L415
https://github.com/NixOS/nixpkgs/blob/2b013e9bc2a74fc63ff6714485b4a4325be59c00/pkgs/stdenv/generic/make-derivation.nix#L281

21:46:24
@tgerbet:matrix.orgtgerbet *

__darwinAllowLocalNetworking for opcache is set to true but I'm not sure if the value is correctly set when we build the extension

https://github.com/NixOS/nixpkgs/blob/8faf826ce5d79eeb731d77927cd850e53d7773d7/pkgs/top-level/php-packages.nix#L415
https://github.com/NixOS/nixpkgs/blob/2b013e9bc2a74fc63ff6714485b4a4325be59c00/pkgs/stdenv/generic/make-derivation.nix#L281

21:47:14
@tgerbet:matrix.orgtgerbetTesting something here: https://github.com/NixOS/nixpkgs/pull/21500421:48:57
7 Feb 2023
@drupol:matrix.orgPol
In reply to @tgerbet:matrix.org
Testing something here: https://github.com/NixOS/nixpkgs/pull/215004
Looks like it works !
07:37:19
@tgerbet:matrix.orgtgerbet Yep, a better way to handle this might exist I suppose but at least it works ¯_(ツ)_/¯ I moved the PR out of draft 09:59:19
@drupol:matrix.orgPolI just validated it 10:00:13
@drupol:matrix.orgPol tgerbet: Is there anything left in your PR ? 18:59:50
@tgerbet:matrix.orgtgerbetI don't think so unless someone comes up with a nicer way to solve this (and I really don't like self merging :) )19:08:02
@drupol:matrix.orgPolI think it's perfectly fine, I don't see what could be changed or improved there.19:23:57
8 Feb 2023
@tgerbet:matrix.orgtgerbetWell it works but I'm still a bit conflicted about introducing something more you need to be aware specific to the PHP extension. The `__darwinAllowLocalNetworking` is already not that easy to discover.18:52:54
@tgerbet:matrix.orgtgerbetAnyway it's merged 😉18:53:13
@tgerbet:matrix.orgtgerbetFor the blacklists.phpt test from what I can see it seems to be flaky. Maybe we could ask for some help in the Darwin channel to see if we can pinpoint the root cause. Otherwise dropping the test on Darwin is likely good enough18:56:29
9 Feb 2023
@drupol:matrix.orgPol I'm trying to convert a PHP developer from Germany to use Nix, and when he does nix shell nixpkgs#php82, Nix starts to compile PHP. 14:44:27
@drupol:matrix.orgPolIt doesn't use the binary cache.14:44:36
@drupol:matrix.orgPolHe's using an Apple M1 (arm64) based computer.14:45:01
@tgerbet:matrix.orgtgerbetWell it's likely due to the issue we have with the tests15:05:34
@drupol:matrix.orgPolOh really?15:32:45
@drupol:matrix.orgPolDamn that's a problem15:33:09
@tgerbet:matrix.orgtgerbetYep, if cannot be built (due to test or otherwise) it cannot be cached16:11:47
@drupol:matrix.orgPolOkay, we need to find out the issue there. It's a bit frustrating to not being able to do that locally !16:14:55

Show newer messages


Back to Room ListRoom Version: 6