Don't serialize IC opcodes

[bnoordhuis]

Translate IC opcodes to their non-IC variants before writing them out. Before this commit they were not byte-swapped properly, breaking the ability to load serialized bytecode containing ICs on systems with different endianness.

cc @chqrlie since it's something you mentioned IIRC

^{edit: note to self: mention in commit log that ICs are recomputed on the fly/as needed}

[bnoordhuis]

I'm going to rework this slightly and read/write the IC atoms as regular atoms in JS_ReadFunctionTag/JS_WriteFunctionTag. Right now they're stored with a distinct tag but that's not necessary anymore.

[chqrlie]

I'm going to rework this slightly and read/write the IC atoms as regular atoms in JS_ReadFunctionTag/JS_WriteFunctionTag. Right now they're stored with a distinct tag but that's not necessary anymore.

Yes. This should remove the need for the 255 73 67 hack in the serializer I suppose. But the bytecode should be scanned after loading to create the IC slots with add_ic_slot1. Unless you make this fully transparent and create the IC slots on the fly in JS_CallInternal, which makes emit_ic useless as well

[bnoordhuis]

Unless you make this fully transparent and create the IC slots on the fly in JS_CallInternal, which makes emit_ic useless as well

Not sure I follow, emit_ic is used in plenty of places. I suppose I could inline add_ic_slot1 into emit_ic because that is its only caller.

[chqrlie]

Unless you make this fully transparent and create the IC slots on the fly in JS_CallInternal, which makes emit_ic useless as well

Not sure I follow, emit_ic is used in plenty of places. I suppose I could inline add_ic_slot1 into emit_ic because that is its only caller.

Well I don't understand the need for emit_ic everywhere in the compiler. Why not create the slots on the fly in add_ic_slot as the properties get actually used in the interpreter loop. This would reduce memory and time overhead for code that is never executed.

[bnoordhuis]

Ah, okay. I don't disagree but that's a separate discussion.