[16:25:21] someone willing to quickly +2 https://gerrit.wikimedia.org/r/c/mediawiki/extensions/SecurePoll/+/827529/ ? [16:26:38] Thanks Lucas_WMDE :) [16:26:59] I was going to ask how CI didn't complain... [16:27:01] shouldn’t CI have caught that? even if we don’t have any tests actually running the code, at least Phan or something like that? [16:27:03] but that's valid syntax xD [16:27:09] oh damn [16:27:13] grr, right [16:27:29] And as most of the parameters are optional... [16:27:35] it won't complain about lack of [16:27:37] how is [] on a constant valid syntax? like, if it were an array sure [16:28:11] it’s “valid” enough that the result is a fatal error at runtime, but not a parse error, I guess? [16:28:19] and no test coverage? [16:28:21] isn't [] the array append operator? how is that valid as a parameter value? [16:28:27] I would've expected phan to have caught it [16:28:42] hmm [16:28:44] s/expected/hoped/ [16:28:46] php -l doesn't like it [16:28:56] does parallel lint for some reason? [16:29:02] hm, in php -a it seems to be a parse error after all [16:41:19] apparently there’s a difference between parsing a function and compiling it? https://github.com/php/php-src/issues/8841 [16:42:03] but Phan should still have caught that tbh [16:42:17] Lucas_WMDE: that seems plausible in php 8+ with the JIT enabled. [16:42:39] php7 didnt have a JIT yet [16:43:13] hm, don’t think I can test PHP 7 locally [16:43:15] * Lucas_WMDE doesn’t really want to try producing a segfault on a production machine [16:48:08] Lucas_WMDE: Toolforge? [16:50:51] good idea ^^ [16:51:12] yeah on PHP 7.3 (toolforge bastion) the second call to f() just throws undefined function [16:51:21] so sounds like JIT is the right idea [16:51:32] but that means we’re back to having no explanation why it’s not a syntax error on PHP 7.2 [17:55:40] mwdebug1001 ;)