In my opinion, the EIP-1283 does not need to be remediated, but rather removed all together. I do understand its intent (making inter-frame communications cheaper), and I proposed an alternative which is cheaper, slightly more powerful, and without the semantic changes to the existing opcodes:
Looking at the history EIP-1283, we see that it has been EIP-1087, formulated in a way which was harder to implement in Parity than in Geth. Then, it was reformulated into EIP-1283, to make it more abstract and less implementation-dependent. However, the semantic complexity was still there, and that caused a consensus issue on Ropsten, which delayed Constantinople once. And now it delays Constantinople once again, due to unintended consequences. Both of these incident can be traced to two things this EIP is doing that were not done before:
- Reducing refund counter (prior to that, refund counter was only increased). That was basis of the Ropsten bug
- Reducing the gas cost on an action. In all other hard forks, the gas cost of operations were only increased.
There is another potential impact of this EIP, which has not been appreciated yet. It will make harder for State Rent group to figure out alternatives to the State Rent, because the edge cases introduced by this EIP will make analysis harder.
My suggestion is to simply exclude this EIP from Constantinople. If the inter-frame communication is still in demand, it can be introduced by a more specialised change, rather than piggy-backing on the existing resource (contract storage). But honestly, I think this particular cost reduction is not a very important feature, and app developers can definitely live without it, while we are concentrating our efforts on Ethereum 1x and scalability improvements instead.