The code below is lifted from /languages/imcc/t/clash.t. I'm trying to understand test writer's intent. Bear with me, please, while I embarass myself:
> ############################## > output_is(<<'CODE', <<'OUT', "defined"); > .sub _test > $P1 = new PerlHash > $I0 = defined $P1 > new P1, .PerlHash > defined I0, P1 > print $I0 > print "\n" > print I0 > print "\n" > end > .end > CODE > 1 > 1 > OUT
1. $I0 (PIR) is set to `defined $P1`;
2. I0 (parrot) is set to `defined P1`. If there is a bug in lifetime analysis, in step 1 above $P1 may mapped to P1. Similarly, $I0 may be mapped to I0;
3. If above-mentioned exists, the two print statements are correct, even if the generated code is incorrect.
Eyebrow raised since file name is 'clash.t', so, in absence of comments about intent, I assume it tries to prove correctness of lifetime clash analysis.
For robustness, a shorter version to test $I*/I* analysis:
I0 = 2 $P1 = new PerlHash $I0 = defined $P1 print I0 print $I0 # should be 21
Or it may be a different clash altogether. Or I may be wrong... Flaviu
