yescallop / fluent-uri-rs

A full-featured URI reference handling library compliant with RFC 3986.

My PGP key: 8771F89F6C5DDDE8080ACD1EE32F586BDAEC7370.

fluent-uri-rs's Issues

Support for IPv6 scoped addresses

As is pointed out by its authors, RFC 6874 was a failure due to several problems with the text. To address these problems, the authors of RFC 6874 have published an Internet-Draft draft-ietf-6man-rfc6874bis-01 that aims to update RFC 3986 and obsolete RFC 6874. Therefore, it should be reasonable that we refrain from implementing RFC 6874 for now and wait until the Internet-Draft is published as an RFC.

Add checked URI resolving

Please add checked URI resolving

Absolute and relative URIs resolving

Hi,

I'd like to ask for a an improvement to fluent_uri::Uri implemenation (as of version 0.2.0-alpha.3 published on crates.io).

There's two methods is_absolute_uri and is_relative_uri. How can I build a new URI based on an absolute and a relative URIs?

for example, I have an absolute URI foo://example.com/path/to/resource and a relative URI other_resource producing a result foo://example.com/path/to/other_resource or with other relative URI ../../another/path producing foo://example.com/another/path?

Builder: query by key-value pairs

I would love to be able to build a query by supplying a series of key-value pairs, and have the builder apply encoding to the keys and values as a browser would when handling a fields in a <form method="GET">.

Elsewhere, the uri_builder crate has a query(&str, &str) method, but I don't believe it handles the encoding.

Taking the idea further, it would be amazing to be able to supply an implementation of serde::Serialize and have that automatically serialised into a query with keys and values encoded. (But I can appreciate fluent-uri might not want to depend on serde, and Serialize is significantly more complicated than an HTML form 😓.)

Remove unnecessary unsafe code in version `0.1`

There is too much unsafe code in version 0.1. It is nearly unmanageable. Yet it is not possible to forbid unsafe code because of Uri<&mut [u8]> and Uri<String>::borrow. At least we should reduce the amount of unsafe code to a minimum.

Stabilize `EStr<Path>::segments`

Thanks for picking up development of this library again!

I'm in a situation where I forced fluent_uri some time ago to implement PartialEq etc. for Uri<_>. With your recent changes I'm almost able to use the upstream repository again.

One thing missing is uri.path().segments() is guarded behind #[cfg(fluent_uri_unstable)]. Do you have any plans on stabilising this or a similar method soon?

Another option might be getting rid of the custom cfg-attribute and putting it behind a unstable feature.

no_std support

Hiya!

From a quick cursory glance over the API of this crate it looks like it would be possible to support no_std without too much trouble, mostly just cfging all std uses with either core or alloc. Is this something you would be interested in? I could of course do so myself but thought I would ask first.

Supporting no_std would allow this crate to be more easily used on more under-powered systems and make it a tad-bit smaller when compiling as well, which would be quite handy in my use-case.

Expose ResolveErrorKind for a developer

ResolveErrorKind is important to decide what error has been occurred before displaying it. Developer should decide how to present it as their might include context, translate as their please and so on.

publish to crates.io

It seems this crate is almost completed, but i can't found it in crates.io and doc.rs, or it is called other name than fluent-uri?