sparsetech/trail

Routing library for Scala and Scala.js

Tim Nieradzik

@tindzk

Indeed, the error message should be: details not specified.

I think it's a good idea to provide the apply() method as you suggested.

I am not sure !# is an intuitive operator. How about this:

val foo = Root / "asdf" / Arg[Int]("foo") / Arg[Boolean]("bar")

(or Index?)

Apart from this it looks really good!

We may also consider providing methods, e.g. Akka provides path() and pathPrefix().

@zarthross I have given you writing permissions to the repository. So if you want to push your changes in a new branch, please go ahead!

Darren Gibson

@zarthross

Awesome. I still have some clean up before I'm happy with it. I'll try and get what I have pushed over the weekend.

Darren Gibson

@zarthross

@tindzk Pushed my idea for a Route DSL to https://github.com/zarthross/MetaRouter/tree/new-route-dsl. I haven't spent much time on the Router portion, but I have some thoughts I'll try to put to code this week.

Tim Nieradzik

@tindzk

Thanks! I'll review it over the next few days.

Tim Nieradzik

@tindzk

@zarthross Impressive work! It's a great idea to use Shapeless. I made a few minor changes here.

Darren Gibson

@zarthross

Kool, I'll look at the changes you made later when i get a chance.

Tim Nieradzik

@tindzk

Do you have an idea how we could merge fill and fillN (see also f455b92aff451799f023d4d9379198ff3248ad17)?

Darren Gibson

@zarthross

Just a cursorer look though, I'm concerned about your move of matches, i think it breaks the type safey of the returned value.

Tim Nieradzik

@tindzk

I actually simplified the type to HList

Darren Gibson

@zarthross

that was the mistake

the Arg Parameter is specifically calculated at compile time based ont eh ROUTE type

Tim Nieradzik

@tindzk

But from what I understood it was untyped before as well.

Darren Gibson

@zarthross

it was type A from CanFillRoute

which was infered from FlatMapper.Aux[ConvertArgs.type, D, A]

Tim Nieradzik

@tindzk

Exactly, and A was determined for fill()'s arguments

But as matches only takes a string, it A is just HList

Am I mistaken?

I had to inline CanFillRoute because IntelliJ didn't find those methods

Darren Gibson

@zarthross

The type A parameter is just the T part of the Args[T] in the ROUTE hlist. I'm pull out the T's to make A

so its more than just an HList, its a specific type known at compile time.

sadly, true taht Intellij didn't like it, but if you inline it you break some of the shapeless checks.

or so i thought

i'll have to check it all later, i'll then take a look at fillN.

Tim Nieradzik

@tindzk

You're right, thanks. I'll try to add the parameters to matches.

Darren Gibson

@zarthross

Trust me, i wouldn't have made teh CanFillRoute extensions if i could have gotten fill and matches to work otherwise...

if you can figure out how to make it work inlined into Route, that would be awesome and i'm curous how to do it.

Tim Nieradzik

@tindzk

My idea was to use Sized.

There is the problem that the wrong fill() function is chosen when only one argument is provided.

With Sized it may be possible to check whether the Product argument is unary.

Darren Gibson

@zarthross

does it not work if you just remove the single argument fill and only use fill() and fill(TP) ?

Tim Nieradzik

@tindzk

I tried that already.

Darren Gibson

@zarthross

so, i'm not really sure how the fillN works in the first place since the function only has one argument but your passing in a list of arguments

Tim Nieradzik

@tindzk

Are there unary tuples?

fillN receives a tuple and creates an HList from it

Darren Gibson

@zarthross

Tuple1 does exist.

Tim Nieradzik

@tindzk

This works because of: fillN((1, 2, 3)) == fillN(1, 2, 3)

Darren Gibson

@zarthross

auto tupled?

Tim Nieradzik

@tindzk

Yes, it's a Scala feature

Darren Gibson

@zarthross

honestly hadn't seen that before... i must have missed it.

just checked it.. kinda surprised, but i like that better.

Tim Nieradzik

@tindzk

https://stackoverflow.com/questions/22385535/converting-single-argument-into-an-hlist-with-shapeless-generic

'The compiler won't automatically wrap a single value in a Tuple1, however—you just get a compiler error.'

That's why I had to override it.

It works fine for some test cases actually, but not for all

Darren Gibson

@zarthross

glad i wrote lots of tests :P

Tim Nieradzik

@tindzk

Yep, it's this one which fails:

        val r = Root / Arg[FooBar]("asdf")
        val i = r.fill(FooBar("dasd"))

Explicitly writing Tuple1(FooBar("...")) fixes it.

Darren Gibson

@zarthross

set T on def fill[T](arg: T) to an upper type bound on product?
def fill[T >: Product](arg: T) maybe?

probably still break foobar since its a Product huh?

leave it to a FooBar to break stuff..

Where communities thrive

People

Repo info

Activity