search

f-o-a-m / purescript-web3

a purescript library for the web3 api
Apache License 2.0
127 stars 24 forks source link

Upgrade foreign-generics #101

Closed martyall closed 6 years ago

martyall commented 6 years ago

This pr upgrades the foreign generics package to be able to use Maybe types, which are available in version 6.0.

martyall commented 6 years ago

Note: we are now using a consistent (de)serialization of the BigNumber type. Before it was possible to use either decimal or hexadecimal encodings, and sometimes there was padding added when there shouldn't be.

Other problems included certain rpc calls that don't exist on all nodes, those now have a Partial constraint.

We are no longer trying to interpret different block structures because the changes here made that more difficult. Also, @XertroV s changes for supporting Kovan blocks were not quite correct -- there are actually many more discrepancies between the two block types and we were just jumbling things up. I think that it's sufficient for the moment to implement a eth_getBlockByHash etc in whatever client library is using a non-standard block structures (in this case parity POA) library, it's easy to follow the pattern laid out in this library or file an issue.

XertroV commented 6 years ago

@blinky3713 - I notice KovanBlock is just in the test module, do we have plans to add Kovan support back to main? Personally I'd like this to be included in the 1.0 milestone since I use Kovan pretty heavily. Happy to do implementation ofc.

On of the issues I foresee regarding custom eth_getBlockByHash implementations is having lots of isKovan type flags and needed to write wrappers around any getBlock calls. Not sure how all that will pan out. That said, Kovan blocks are technically different blocks so it's probably worth keeping different types or implementing kludges (e.g. a KovanBlock -> Block method) a per-project thing.

One idea might be to allow a function to be injected in the block processing thing? You could use an AVar within Web3 to store the function and allow it to be updated, though this might be bad design (e.g. if someone else is working on my code it might not be apparent that the behaviour of getBlock now doesn't match the ps-web3 docs)

Hmm.