I think the idea was to make OP_RETURN only and OP_RETURN {pushdata} standard, but reading the code at script.cpp I think than any opcode after OP_RETURN would be accepted as standard.
I don't think this has any security implication, nor if it's worth the trouble to change that. Maybe the solution is just to document it correctly.
Could you explain in more detail?
The IsStandard() test uses the usual template matching code - I don't see how that could accept anything but the template.
Note that there are two separate things here:
the code reads: else if (opcode2 == OP_SMALLDATA) { // small pushdata, <= 80 bytes if (vch1.size() > 80) break; }
Where it should read something like :
else if (opcode2 == OP_SMALLDATA) { if (opcode1 <= OP_16) { // small pushdata, <= 80 bytes if (vch1.size() > 80) break; } else break; }
It'd probably make sense to just make the test be:
if (vch1.size() > 80 || vch1.size() < 1)
break;
That's why OP_PUBKEY and OP_PUBKEYHASH can't match an arbitrary opcode, and there's no good reason to include an empty data string - use a plain OP_RETURN scriptPubKey instead.
What is the problem with OP_RETURN OP_CHECKSIG (or OP_RETURN OP_ANYTHING) being standard?
I vote we just document it as "that is the way it is."
@gavinandresen I'm fine with that - might help discourage people from over-thinking what's supposed to be in there...
Closing, consensus is "fine the way it is." If anybody feels strongly about more documentation, submit a pull request to https://bitcoin.org/en/developer-guide#standard-transactions (but I don't think it is worth mentioning, this is a weird nobody-is-going-to-do-it-that-way edge case).