jjcnn
jjcnn
@valeryz : This issue is blocking for two open PRs in the scilla repo. The PRs aren't critical, but we'd like them to be merged because they will be needed...
The check has been moved to `ScillaParser.mly` to make `scilla-checker` able to check for the problem. The issue also affects map value types at the moment because of issue #261.
Note that the design for this issue may affect the design of issue #516.
Consider adding iterators for maps as well. This would definitely impact the design of #516.
I thought about this a while ago, and I couldn't come up with a better way to do it than what @vaivaswatha suggests above, so maybe we should just retire...
Can we guarantee the traversal order? i suppose we can, since we appear to be able to guarantee determinism of `Hashtable.to_list` or whatever it's called in OCaml.
Address types also need fixing.
I am changing `ScillaParser.mly` to disallow function types as map value types. The production `t_map_value` should be removed when this issue is fixed.
Note that there is a small chance that this will cause existing contracts to no longer parse, so do check existing mainnet contracts before deploying, or postpone until the next...
It is also necessary to re-enable pretty-printing in the hash_lib function. It has been disabled because staging prevents us from printing s-expressions directly.