Tim Düsterhus

> The following test script also appears to log the incorrect information: The cause probably is that you can't naively `memcpy` the zval, because the reference counting of strings is...

I'm still seeing a leak: ``` b57e622520ad# ../caddy/frankenphp/frankenphp php-server --worker worker-getopt.php 2024/05/22 13:44:51.696 WARN admin admin endpoint disabled 2024/05/22 13:44:51.696 WARN http.auto_https server is listening only on the HTTP port,...

And the test script with the `file_put_contents` from yesterday’s review still outputs the wrong data for the `after:`.

@dunglas For the test script in https://github.com/dunglas/frankenphp/pull/796#pullrequestreview-2067722841 I'm still seeing the wrong data being written to the `log` file: ``` before: worker-getopt.php before: worker-getopt.php before: worker-getopt.php before: worker-getopt.php before: worker-getopt.php...

> In the last commit, I fixed the crash without changing the existing behavior (the docs change is now irrelevant, I'll revert that). I think it's good enough for now....

> I think I've found a way to keep the userland and internal structures in sync but I didn't create a test case. I can confirm that internal functions called...

@dunglas Can confirm with the cli-server. Sorry for the confusion. This looks good to me then!

> I also believe that Frankenphp uses Semver, which php's `version_compare` doesn't really understand and usually starts with a `v`. In that case, you only need to check the `major`...

> and I've no idea from where this function gets the number It's taken from the `zend_module_entry`: https://github.com/dunglas/frankenphp/blob/469070ce8573fc2cd9453d2559d8b5d9d0fa93fb/frankenphp.c#L419

@SoftCreatR In the reproducer steps your 2 alternatives are identical.