Alastair Robertson

Well the debug output all looks good as far as I can tell, so I'm running out of ideas. Unless there are any other tracepoints this happens on, we might...

I'd rather we just fix it properly in all situations, and the easiest way to do that is (IMO) to fully remove the multi-delete syntax. It's pretty new (only appeared...

Is this what's being suggested? ``` @a = 1; @b[0] = 0; @c[0] = 0; delete(@a, @b[0]); // multi-delete delete(@c, @b[0]); // new delete syntax ``` I think that this...

I think I might have misunderstood the discussion above. The staggered release suggestion was to ensure we could differentiate between the multi-delete and the new syntax, but I think the...

Codes of conduct are used in lots of projects, but there are also plenty of projects larger than us which don't use them. [A search on Hacker News](https://hn.algolia.com/?dateRange=all&page=0&prefix=true&query=code%20of%20conduct&sort=byPopularity&type=story) shows that...

Anyone have any ideas for a better way to test this particular change?

Updated to mark a couple of extra helpers as impure (`get_ns_current_pid_tgid` and `get_stack`) as they write their results to a provided pointer parameter. Tests are passing now!

Added an expanded comment about the dangers of marking helpers as pure

I was wondering about them myself, but in what circumstances can that memory change within the context of a single BPF program?

@wugeng617 Could you paste the full output of `bpftrace --info` please? Like @jordalgo, bpftrace 0.23.3 is producing correct-looking output on my machine. v0.22 did make significant changes to `buf`'s implementation...