operands are out of order

[ThinkOpenly]

I made an unfortunate choice long ago to pull operands from the "function signature" of the function clause execute instances. The order of the inputs there need not match the order of the operands in the assembly syntax, and indeed there might be inputs which are not operands. For example, the vaadd.vv instruction's assembly syntax is:

vaadd.vv vd,vs2,vs1[,vm]

However, the function clause execute signature looks like:

function clause execute(MVVTYPE(funct6, vm, vs2, vs1, vd))

Here, there is an extra input, "funct6", and the operands are in fairly random order. This is also true of the signature for the other stanzas, including mapping clause encdec and mapping clause assembly.

The operands, not surprisingly, are well represented in the actual assembly syntax in mapping clause assembly:

mapping clause assembly = MVVTYPE(funct6, vm, vs2, vs1, vd)
  <-> mvvtype_mnemonic(funct6) ^ spc() ^ vreg_name(vd) ^ sep() ^ vreg_name(vs2) ^ sep() ^ vreg_name(vs1) ^ maybe_vmask(vm)

The latter line there translates fairly easily into vaadd.vv vd,vs2,vs1[,vm].

I think we need to move the operand extraction from the function clause execute processing to the mapping clause assembly processing, if possible.

[Shubhf]

Hey Can I take up this issue?

[ThinkOpenly]

Hey Can I take up this issue?

@Shubhf Thanks for the offer. @Linda-Njau has been working on this, and it quite close, in #34 .

[ThinkOpenly]

Fixed by #34, that is now merged.

Closing...