Re^2: Where was I called?

[Ovid]

That's one thought, but it turns out there's a bug in the scheme.

while ( my $value = some_sub() ) {  # line 50
    my $next_value = some_sub();    # line 51
}

This is something I forgot about, but it's very, very irritating. The first time some_sub() is called, caller will correctly report that line 50 is the calling line and the second time some_sub() is called, it reports line 51. However, the second time through the while loop, the first some_sub() is reported as being on line 51! That totally blows my initial scheme.

[diotalevi]

Wow, thanks Ovid! I never realized that problem existed. Just to explain, whenever a nextstate op is seen the global variable PL_curcop is set and that's some of what caller() uses. You know line 50 as you enter the loop, then the first line in the loop is 51 but you never reach that outer nextstate op to say you're testing a condition on line 50 again. I bet if you were really interested in solving this problem in a JFDI way that you could change that "goto 9" to be "goto 2" or promote the conditional expression into a do{} block. But that way lies madness so you won't do that.

### caller() will report line 50.
2  <;> nextstate(main 2 x:50) v
3  <{> enterloop(next->8 last->d redo->4) v
9  <0> pushmark s
a  <#> gv[*X] s/EARLYCV
b  <1> entersub[t2] sKS/TARG,1
c  <|> and(other->4) vK/1
       ### caller() will report line 51
4      <;> nextstate(main 1 x:51) v
5      <0> pushmark s
6      <#> gv[*Y] s/EARLYCV
7      <1> entersub[t4] vKS/TARG,1
8      <0> unstack v
           goto 9
d  <2> leaveloop vK/2
e  <@> leave[1 ref] vKP/REFC

[Ovid]

You know, that almost made sense to me. OK, no it didn't. I have no idea how to read that output :) The reason why caller gets it wrong now seems to make sense, though, so thanks!

[ysth]

That's B::Concise output - see its doc.