Re: [PATCH][gdb/testsuite] Support .debug_aranges in dwarf assembly

[Simon Marchi <simon.marchi@polymtl.ca>]

On 2021-08-27 1:10 p.m., Tom Tromey wrote:
> Simon> But doing it this way makes it so that you can only invoke arange when
> Simon> you are in aranges' body, isn't that useful?  I guess the downside to
> Simon> redefining the proc everytime is performance, but that's really not a
> Simon> concern here (it runs quickly enough).
> 
>>> To do that, it also have to delete the 'arange' proc after evaluating
>>> the body.  I suppose that would be alright by me.
> 
> Simon> Really?  This technique is used in proc rnglists, and that doesn't seem
> Simon> to cause a problem.
> 
> AFAIK Tcl doesn't have any kind of lexical scoping for procs.
> So, after "proc arange" is evaluated, the binding stays around.
> 
> This contradicts the what you were saying: "you can only invoke arange
> when you are in aranges' body".  I think that's not the case, you can
> invoke arange any time after any aranges call in the current invocation
> of runtest.

Ah, I didn't know, I said that without actually trying it.  I assumed it
worked like all other languages :).

> I'm not concerned about the performance, I guess.  It's just
> un-idiomatic to define a proc in a proc, normally this would only be
> used for tricky things like changing a proc body at runtime.  So when I
> see this sort of thing, I start looking for the trick.  There's nothing
> incorrect, it's just less clear than it could be.

So, defining the proc inside another is not useful then.

Simon