Robert

It seems reasonable to me that `unknown` parameters should be in text format (want to link some documentation that supports this but am coming up short...). From my point of...

A potential future solution for this might be via ghcup release channels, as suggested in #108 (but that would require GHC CI to generate a ghcup metadata file at a...

Meanwhile postgrest has migrated from configurator-ng to configurator-pg, so from that point of view this update is no longer necessary.

With the changes from #93, critbit is closer to be stackageable. Only missing thing is the dependency `bytestring-trie`, which in turn could be added directly.

Here's a relevant stackoverflow question: https://stackoverflow.com/questions/53456798/elm-attribute-onerror-adds-data-onerror-attribute-instead

For my specific use case, ports should work fine as shown in https://github.com/mpizenberg/elm-pointer-events/tree/4.0.0/examples/Pointer. I'll update the title to ask for this behaviour to be documented.

(Marking read for review to see if that triggers CI.)

A few notes: * I can reproduce the problem, but I had to add another zero to the large number. * The analogous code for [elm-community/parser-combinators](http://package.elm-lang.org/packages/elm-community/parser-combinators/2.0.0/Combine) crashes similarly. * You...

@ddd356 That's because you're not parsing the URL in your last example. `setRequestQueryString` encodes the query parameters properly: ``` > getUri $ setQueryString [("foo", Just "[]")] defaultRequest http://localhost/?foo=%5B%5D ``` The...

It seems like we *should* be running an http server ourselves in the tests instead of relying on an external service. (Just ran into this myself trying to test https://github.com/snoyberg/http-client/pull/477.)...