Skip to content

Conversation

@cdonovick
Copy link
Owner

@cdonovick cdonovick commented Aug 2, 2019

Sorry, ignore this comment

@cdonovick cdonovick requested review from phanrahan and rdaly525 August 2, 2019 17:27
@phanrahan
Copy link
Collaborator

LGTM. Thanks for the quick turnaround.

@phanrahan
Copy link
Collaborator

Shouldn't we also change encode ...

def encode(inst, *args):
    #print(inst, type(inst))
    if isinstance(inst, AbstractBitVector):
        word = int(inst) << inst.bitfield
    elif isinstance(inst,Enum):
        word = inst.value << inst.bitfield
    elif isinstance(inst,Product):
        word = 0
        for key in type(inst).field_dict.keys():
            word |= encode(getattr(inst, key))
    elif isinstance(inst,Sum):
        t = type(inst)
        if hasattr(t, 'tags'):
            i = t.tags[type(inst.value)]
        else:
            i = list(t.fields).index(type(inst.value))
        word = (i << t.bitfield) | encode(inst.value)
...

@cdonovick
Copy link
Owner Author

Probably, I don't use encode anywhere, feel free to modify as you see fit.

In the long term I think it would be an appropriate to remove encode and build anAssembler with the same logic. The AssembledADT infrastructure that is about to exist relies on people subclassing AbstractAssembler so that it can translate adt[key] and adt.attr to bv[idx]

@phanrahan
Copy link
Collaborator

Let me build out encode so that is assembles the pdp8 and riscv, and then we can figure out how to merge those ideas into Assembler.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants