Brainstorm Uri Representation

[hyperium]

Brainstorm Uri Representation

[seanmonstar]

As the docs for variants point out, a target server should only need the
AbsolutePath. The others are used mostly by proxies.
http://seanmonstar.github.io/hyper/hyper/uri/type.RequestUri.html
On Sep 17, 2014 12:19 AM, "Jonathan Reem" [email protected] wrote:

I've found the current version unergonomic to work with in downstream
libraries without just ignoring certain variants. Not sure right now what's
better, but I'll think about it.

—
Reply to this email directly or view it on GitHub
#39.

[reem]

That's what we do in Iron. I guess that's really the best we can do.

[seanmonstar]

Though, we could add some convenience methods to get to the path.

Also, I've been thing that it might be cool to have a stronger typed value
than just a String in there...
On Sep 17, 2014 2:53 PM, "Jonathan Reem" [email protected] wrote:

That's what we do in Iron. I guess that's really the best we can do.

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

[reem]

We should probably use rust-url and hold a Url value instead of a String.

[reem]

The tricky thing is that rust-url does not support parsing from only a relative url - you have to give it a full absolute path.

We have a huge hack in Iron where we construct a url for rust-url to parse, but I think that won't be necessary if we have all the info inside hyper.

[seanmonstar]

Yea I wanted to use a Path from rust-url, but there isn't one yet. I tried
asking for one in servo/rust-url#14, but since I
could only offer hypotheticals, it was closed. Maybe you could give
concrete needs for Iron.

[reem]

It looks like it was closed with a fix, does the method he added not do what we want?

[seanmonstar]

I guess I missed that. I'd need to look into it. It looks like it validates that the value is a proper path, but otherwise returns a String.

Would it be doing too much if hyper transformed a Path + Query into something, or perhaps having a Decodable fn for getting a query? Perhaps that's better left to library land.

[reem]

I think it's fair if Hyper provides a typed representation for request-line, with precedent being typed headers and other such things. We should just make sure to not lose any information.