|
02:02
jnap joined
02:32
FROGGS_ joined
02:45
jnap joined
03:34
jnap joined
09:10
colomon joined
09:55
tgt joined
12:23
mj41 joined
12:37
FROGGS joined
15:25
benabik joined
16:41
shipwright joined
19:23
tgt joined
19:25
pmurias joined
|
|||
| FROGGS | jnthn: there seems to be a bug that triggers "Cannot invoke null object" | 20:08 | |
| it happens when doing the following | |||
| 1) module Foo has a my scoped hash with an element and an our sub that just return the value of that element by key | 20:09 | ||
| 2) module Bar uses Foo | |||
| 3) you use Bar | |||
| the problem only appears when the modules are precompiled | 20:10 | ||
| and when it is about a hash, a my scoped variable just works fine | 20:11 | ||
| ohh, an array exposes this bug also, when return element 0 for example | 20:12 | ||
| here code is NULL: MVMObject * MVM_frame_find_invokee(MVMThreadContext *tc, MVMObject *code, MVMCallsite **tweak_cs) | |||
| the problem seems to be within the precompiled module Bar | 20:14 | ||
| ahh, I forgot to mention, Bar call that our scopes sub | 20:15 | ||
| and now I wonder what it is about... | 20:16 | ||
| is the invoke_o where the second arg is NULL about call the our scoped sub? or is that something within that sub? | |||
| like calling atkey/atpos | 20:17 | ||
| okay, it seems to be about atkey/atpos, because return the array just works | 20:23 | ||
| hmmm, using nqp::atkey instead doing it perl6ish works... | 20:25 | ||
|
20:32
tgt joined
20:35
benabik joined
21:39
tgt joined
|
|||