Bob Weinand

@mabdinur Yep, needs a change in the test-agent to filter them there.

I think you should test `datadog.appsec.sca_enabled=1` in one of the tests instead of using the ENV there. I believe it needs to be handled in configuration.c so that it's parsed...

This approach is problematic because of https://github.com/php/php-src/issues/10335#issuecomment-2066427171. Maybe we can use it once it's somehow fixed upstream?

@amberovsky Is there a possibility that you could get core dumps and post us at least the stacktraces? That would be probably helpful in triaging. Thanks!

Hm, the crashing thread is thread #1 and it's full of question marks. There are three options: a) you don't have debug symbols for PHP itself b) it's JIT code...

Just a quick note: This test tests behaviour under a sigkill (it's 9, not 11), just the matching in the diff doesn't match the behaviour observed, i.e. the system doesn't...

I also think dogstatsd would be a good fit for a small crate in libdatadog. Can we move it there? :-)

Hey @deiga, while we provide https://packagist.org/packages/datadog/dd-trace (which you are free to include if you want IDE autocompletion for example), composer packages can't provide extensions themselves, only dependencies on extensions, hence...

Hey @rares-mollie, You are right that you can't get hold of already closed spans. However, you could add a hook to e.g. `mysqli::query` yourself, where you can access that span...

I actually meant using `trace_method`, not `hook_method` - if a method is hooked via trace_method() multiple times, it will still share the same span parameter across all callbacks.