chore: Upgrade to hugr 0.8.0 #66
Conversation
| @@ -201,7 +200,7 @@ compileConst parent tm ty = do | |||
| constId <- addNode "Const" (OpConst (ConstOp parent (valFromSimple tm))) | |||
| loadId <- case ty of | |||
| HTFunc poly@(PolyFuncType [] _) -> | |||
| addNode "LoadFunction" (OpLoadFunction (LoadFunctionOp parent poly [] (FunctionType [] [HTFunc poly]))) | |||
| addNode "LoadFunction" (OpLoadFunction (LoadFunctionOp parent poly [] (FunctionType [] [HTFunc poly] []))) | |||
There was a problem hiding this comment.
Why are the extension reqs for LoadFunction empty?
There was a problem hiding this comment.
LoadFunction is only used for const functions i thought? The extensions in the function that's loaded are boxed up until it is called
There was a problem hiding this comment.
I have no idea, just checking 😅
But that makes sense 👍
brat/Brat/Compile/Hugr.hs
Outdated
| @@ -295,13 +294,14 @@ compileClauses parent ins ((matchData, rhs) :| clauses) = do | |||
| didntMatch outTys parent ins = case nonEmpty clauses of | |||
| Just clauses -> compileClauses parent ins clauses | |||
| -- If there are no more clauses left to test, then the Hugr panics | |||
| Nothing -> let sig = FunctionType (snd <$> ins) outTys in | |||
| addNodeWithInputs "Panic" (OpCustom (CustomOp parent "brat" "panic" sig [])) ins outTys | |||
| -- TODO: This should just be [BRAT] - would that work? | |||
There was a problem hiding this comment.
Don't ops add their own extension as a requirement implicitly?
There was a problem hiding this comment.
Not as part of the signature - I've implemented this comment anyway
| @@ -437,7 +438,7 @@ compileWithInputs parent name = gets compiled >>= (\case | |||
| _ -> compileWithInputs parent outNode >>= \case | |||
| Just calleeId -> do | |||
| callerId <- addNode ("indirect_call(" ++ show calleeId ++ ")") | |||
| (OpCallIndirect (CallIndirectOp parent (FunctionType ins outs))) | |||
| (OpCallIndirect (CallIndirectOp parent (FunctionType ins outs bratExts {-[]-}))) | |||
There was a problem hiding this comment.
Should this be empty or not, c.f. LoadFunction?
There was a problem hiding this comment.
The extensions here should match the extensions of the callee, but we're overapproximating
brat/Brat/Compile/Hugr.hs
Outdated
| -- HELP! This doesn't work at all when extensions are involved!!! | ||
| -- |
There was a problem hiding this comment.
Still looking for help? What's going wrong?
There was a problem hiding this comment.
This is only a problem when we stop overapproximating extensions, I'll zap the comment
brat/Data/Hugr.hs
Outdated
| bratExts = | ||
| ["prelude" | ||
| ,"arithmetic.int_ops" | ||
| ,"arithmetic.float_ops" | ||
| ,"collections" | ||
| ,"logic" | ||
| ,"tket2.quantum" | ||
| ,"BRAT"] |
There was a problem hiding this comment.
arithmetic.int_types and arithmetic.float_types are missing
brat/Data/Hugr.hs
Outdated
| args = [funcToSeq outerSig, TASequence (funcToSeq <$> innerSigs)] | ||
|
|
||
| funcToSeq (FunctionType ins outs) = TASequence [toSeq ins, toSeq outs] | ||
| sig = FunctionType (toFunc <$> (outerSig : innerSigs)) [toFunc outerSig] ["BRAT"] |
There was a problem hiding this comment.
Do we also need to include the reqs of the input graphs?
Resolves #63