feature/void #472
feature/void #472
Conversation
There was a problem hiding this comment.
I think I would prefer a different option: introducing a dedicated Discard (or so) JSOperation, similar to e.g. TypeOf. You can probably more or less copy the code related to TypeOf to get it to work.
The reason is that I don't think we would want to generate many void operations ourselves in our code generation pipeline since they are pretty uninteresting. So this we, we avoid "diluting" the other unary opertors.
I think this feature would be mostly useful to support compilation of more .js files to FuzzIL. As such, could you also add support for it in our Compiler? You can take a look at issue #437 or some recent PRs to see how those changes work. Also again, you can probably check how we handle TypeOf and deal with void in a similar manner.
Thanks! :)
| @@ -664,6 +664,8 @@ public struct JSTyper: Analyzer { | |||
| set(instr.output, .boolean) | |||
| case .BitwiseNot: | |||
| set(instr.output, maybeBigIntOr(.integer)) | |||
| case .Void: | |||
| set(instr.output, .anything) | |||
There was a problem hiding this comment.
This needs to be .undefined I think (the output of the void operator is always undefined IIRC)
|
Nice! This looks good, I'm just wondering what happened to this commit for the compiler support bb872e1 @vi3tL0u1s? We can also just land this CL as it is and add the compiler support in a follow-up PR. WDYT? |
|
Hi @saelo, I’ve added the compiler support as you suggested. Thank you so much for the reminder! |
|
Ok great, thanks! I think there's a merge conflict though, and also I think there used to be a compiler test as well (bb872e1#diff-e3c74f57e01c1d64375830a153bdede28b838e44615cad0ba7085d5754390c2e), can you add that again, too? |
|
Hi @saelo, I’ve added the compiler test. Regarding the conflict, I believe it might be due to this PR not including the logic for the delete feature, which was merged earlier. I’m not entirely sure how to resolve this. Could you provide some guidance? |
|
Ah yes I see, let me try to figure out how to resolve that :) |
|
Ok, that worked. I guess I'll just merge these commits, then locally squash the merge commit, then force push. A bit hacky but oh well :) |
|
Okay done! I had to change the test a little bit because it ends up redefining the |
|
Thank you so much @saelo! |
Add a new UnaryOperator
void