Decoders are ugly about rho

[arouel]

+1

[bryce-anderson]

Do we know of a symbol that means 'decoding'? I would like to do something as follows

GET / "foo" |#>[JValue] { result: JValue => ??? }

but this restricts us to having a type per decoder unless we want to place it at the end ala.

GET / "foo" |#> { result: JValue => ??? }(dec: Decoder[JValue]) // type annotation for clarity

Of course this could be solved with an alias,

GET / "foo" |##>(dec: Decoder[JValue]) { result: JValue => ??? }

This formulation will remove the ability to add more header requirements or even another codec after adding a codec. The second one isn't really a deal breaker as you will likely want to pack them together beforehand anyway.
I'm not sure of this yet, but it could allow for the removal of the CodecRouter type.

[arouel]

I've no better symbol yet for decoding but I like your choice. I like your second example and the alias for it, that's what I would use.

[bryce-anderson]

Commit a45f65e did away with our own decoder type in favor of the http4s EntityDecoders.

[arouel]

Nice that we use the http4s EntityDecoders now.

[bryce-anderson]

Yes, and hopefully we can coalesce more of these niceties into shared code. That is part of what rho is for: find the squeaks in http4s. I would like to find a way to share the ResponseGenerators that the http4s.Status currently uses, but I would really like to get the response type information.

[arouel]

Awesome, currently I'm really busy and could not help much but the direction sounds really good. Thank you

[bryce-anderson]

Glad you like the direction. Hope your load lightens soon: its nice when
you're around pushing those high quality commits and calling me out on my
poor test/documentation discipline. Glad you're on the team.

On Tue, Aug 5, 2014 at 3:57 PM, André Rouél [email protected]
wrote:

Awesome, currently I'm really busy and could not help much but the
direction sounds really good. Thank you

—
Reply to this email directly or view it on GitHub
#15 (comment).

[bryce-anderson]

I think this should have been closed.