sysapps / app-uri Goto Github PK

Section 7 - Point 3 reads:
The developer should not be bother as to wether they are using http:// or app:// - the Web's capabilities and APIs need to just work!

Should be:

The developer should not be bothered as to whether they are using http:// or app:// - the Web's capabilities and APIs need to just work!

A test suite should be created, to allow spec verification and to move it to LC.

Pass pub rules, etc.

The spec is currently lacking security considerations about content type sniffing and being careful with dealing with packages.

The spec should be reviewed to make sure we didn't miss any additional privacy considerations.

When dereferencing, if the UA hits a 500s or 404's must return an empty response body.

Based on: https://bugzilla.mozilla.org/show_bug.cgi?id=827243

http://www.w3.org/2012/sysapps/app-uri/

The spec is missing a link to the Web Origin Concept spec where it talks about "origin".

Giridhar wrote:

I would recommend under the Security Considerations section that the text be modified to be "The user agent needs to make sure that a symbolic link (or similar) inside a package does not break out of the package and end up pointing to a physical file on the end-users device." I also don't believe everyone has a common understanding of what "break out of the package" means in this context, so it would be good to clarify this phrase.

The UUID can be used to track people. It should be possible to regenerate it when clearing private data.

Spec should contain a section describing the use cases and requirements.

http://lists.w3.org/Archives/Public/public-sysapps/2013May/0045.html

What happens to request headers when, for instance, XHR is used to get a local file.

Elsewhere, Tatsuya Igarashi proposed:

Sony has investigated a feasibility to playback media files within a packaged web application, especially by using Media Source Extensions(MSE) [2] and Encrypted Media Extensions(EME) [3] for offline playback. The App URL scheme is required to use to specify a media file in a package. In this time, we have tested the media playback with File URL of WebView instead of App URL. And we have encountered an issue that HTTP range header does not work correctly to access local file resources. It may be a bug of the web runtime implementation. But it is the background that we think of this clarification. Actually, it would not be noticed that HTTP Range has to be supported for MPEG DASH [4] On demand profile by using MSE.

I suppose that most working group members assume that XHR with “The App URL scheme” fully follows HTTP RFC-2616 [5] and there is no problem with media playback. But it would be nice to clarify that AppURL can be used for video/audio playback as well as other resources, e.g. images. Because media playback is crucial for packaged web applications.

I suggest to add the following notes to the two examples in the “The: App URL Scheme” spec. Also your suggest for this clarification is welcome.

The mention of UUID in the text preceding Section 6.1 implies that it is required, when Sec. 6.2 makes it clear that it is recommended.

Would be good to layer this URI on top of fetch.

Detailed here:
http://www.w3.org/2013/08/27-sysapps-minutes.html#action01

The spec says:

The identifier represented by the authority is bound to an instance of an application for the life of that application instance: that is, until that instance is destroyed (e.g., the application is uninstalled from an end-user's device).

How long the authority is bound to a particular instance should be defined by the runtime spec. The runtime spec needs to define the life of the authority (e.g., if it's reconstructed when private data is cleared, when the app is reinstalled, etc.).

See: https://twitter.com/roessler/status/472119915961262080

Giri wrote:

The ABNF spec does not place any requirements on use of unique identifiers. I would suggest changing " And, and uuid is defined in [UUID]" to "See Section 6.2 for more information on the use of UUID in construction of the authority field."

I see from #24 that app didn't make it into IANA's URI schemes, even as a provisional registration.

http://www.w3.org/TR/2015/NOTE-app-uri-20150723/ now sadly says:

This specification is being republished as a Working Group Note as an indication that it not being progressed further as a Recommendation track document.

I guess this follows from closing of the sysapps WG

Does this mean that app URI scheme is now defunct, or is it something others might be thinking about moving forwards?

We had - perhaps prematurely - used app: as part of Research Object Bundle identifiers - generating a unique identifier for a resource within a structured ZIP file.

Are there other de-facto users or historical users of app:// URIs you know about?

Would it make sense now to register app with IANA as a historical scheme? I can be the individual.

Currently, there is no conformance requirement on UA's in section "Query and Fragment components". There needs to be at least one MUST in there to allow for the creation of tests.

Received the following feedback offlist.

I think some readers will not understand the examples[1] because they have to be able to figure out what the window.location has been set to in order to understand why the tests for equality are true. If they could do that, they wouldn't need the examples at all because they'd already understand how the "app: uri" works.

The first sentence in the example is: "The following example shows [HTML]'s window.location using then app: URL." There's an editing error at "then". I'm guessing something was dropped accidentally.

It needs to be something like:

The following examples shows the results of using "app://com.foo.bar/index.html#example" for [HTML]'s window.location

The rest of the examples make sense only if they relate to some particular URL. Otherwise, they're just showing unrelated features. One of the examples has "img.src === "app://c13c6f30/example.gif"" so must have a window.location different from example 1. They all should be related to a single URL that is given at the top.

http://www.w3.org/2012/sysapps/app-uri/

Need to begin the process of registering the URI scheme with IANA and IETF.
http://tools.ietf.org/html/rfc4395

Now that we have a preliminary test suite, the spec needs to link to it.

In IRI spec domain is case insensitive and I'm wondering if in our case we should treat window.location.host as domain or as path?

Generally when getting files from package using XHR host (i.e. f15a6d20-cefa-13e5-1972-800e20d19a76), host functionally to me is more like domain (case sensitive) than path.

If we should treat host case sensitive then we should mention that in specification explicitly since as it is now I expect app scheme will work similarly to http, and it that case host should be case insensitive.

What do you think?