Kévin Dunglas
Kévin Dunglas
If this is a thread safety issue, as a workaround, maybe could we override `opcache_invalidate()` to add some locking mechanism?
For the `opcache_reset()` problem, couldn't we replace the original function by a new one wrapping it around a global lock that pauses all incoming requests? This should prevent the issue...
Maybe could we wait for all current PHP requests to finish, flush the cache, then unblock buffered requests. This is not really optimal, but at least it will prevent the...
Couldn't this be implemented userland with the new `frankenphp_log()` function?
First option is definitely the best!
I would rather update the recipe. If we do that, we'll recommend installing outdated MySQL versions that may have known security vulnerabilities.
Please no PSR-7, immutable objects are not even compatible with the HTTP/2(+) spec where everything is a stream.
Can you use the long form of options for clarity please?
I use a Mac but don't manage to reproduce locally.
IMHO we should consider this a bug and fix it, even if there is a small performance penalty. Maybe can we provide an option to opt-out, but compatibility with normal...