Alastair Robertson

> I think if folks want to make a case for replacing these prefixes with keywords to make them more understandable to new users, that's a separate discussion Another related...

I'd still go with a keyword, for the extensibility reasons previously discussed, but I also think this point raised by @viktormalik is actually pretty important: > ### 2. Declarations with...

What's the main benefit of having no keyword? Terseness, or something else I'm missing? If it's just for conciseness, then I think we need to think about the scenarios variable...

Preventing nested tuples from being used as map keys seems a reasonable first step, but I really do think we need to make strings in tuples work transparently, regardless of...

> > Should this Scope node be called Block instead? > > I went back and forth on this and TBH I'm not sure what the correct answer is. Scope...

That could just become this, no? ``` void SemanticAnalyser::visit(While &while_block) { if (is_final_pass() && !bpftrace_.feature_->has_loop()) { LOG(WARNING, while_block.loc, out_) accept(*this); loop_depth_++; while_block.scope->accept(*this); loop_depth_--; } ``` I don't think there's any...

> I think the condition _should_ exist within the scope of the loop's body. Let's look at a `for` loop for example: Agreed about for-loops, and that's why I think...

> @ajor I looked into this but I don't think there is any additional optimization here. At the time of scope iteration, we're only already iterating over eligible scopes and...

> For adding multiple configs, what about comma separated (feels more similar to the colon separated probe syntax) ? There's also the third option of using semicolons to match the...

> How do we plan on thinking about probe args (colon separated) vs probe attributes? When do we pick one over the other when adding or extending probes? I'd say...