pushState + hash on root URL

[tkhyn]

Hi (and big thanks already for all the great work on Aurelia),

I've started to play around with the framework by setting up an app which router is configured with config.options.pushState = true; to use pretty URLs.

I would still like to be able to use bookmarks on my root page (such as http://host/#bookmark), and I noticed that the location was systematically turned into http://host/bookmark. # is however not stripped when using an URL other than the root location (i.e. http://host/whatever#bookmark works fine).

This seems to happen here.

Is it the expected behavior to mutually exclude bookmarks or routing on the root URL when using pushState = true ? If that's the case I'll challenge that, I think!

[EisenbergEffect]

I think it's likely just a use case we didn't initially consider. @bryanrsmith can comment on this more though.

[tkhyn]

Having a closer look at the source code, I set config.options.hashChange = false; and it works perfectly.

So the only thing that'd be missing here is a mention of hashChange somewhere in docs (knowing of course that it's a work in progress).

Or I may suggest something that could be more sensible (but I don't know all the in and outs and it may break other things) :

diff --git a/src/index.js b/src/index.js
index aee7c41..24f20e2 100644
--- a/src/index.js
+++ b/src/index.js
@@ -51,8 +51,8 @@ export class BrowserHistory extends History {
     // Normalize root to always include a leading and trailing slash.
     this.root = ('/' + this.options.root + '/').replace(rootStripper, '/');

-    this._wantsHashChange = this.options.hashChange !== false;
     this._hasPushState = !!(this.options.pushState && this.history && this.history.pushState);
+    this._wantsHashChange = (this.options.hashChange === undefined ? !this._hasPushState : !!this.options.hashChange)

     // Determine how we check the URL state.
     let eventName;

(as I'm still pretty new to Aurelia / Karma / jspm and even Node.js, I struggled to find a reliable way to test the activate() method. Is there an example somewhere in an Aurelia module where something similar is done?)

[bryanrsmith]

I think this is a duplicate of aurelia/router#225. As far as I know it was indeed an oversight. I agree with your proposal, @tkhyn, to change the default behavior so that the router must be explicitly configured to rewrite incoming hashes as paths.

This is technically a breaking change for applications that want incoming hash-based fragments to be transformed, but it seems unlikely to affect many users (see http://caniuse.com/#feat=history), and the migration is trivial.

[bryanrsmith]

@tkhyn I'm not sure of a good reference for testing the activate method-- they're a bit lacking at the moment. Maybe some of the tests in aurelia/router? Unfortunately you'll probably need to just read the code and build tests around it. This would be a very welcome contribution, if you're up for it! Feel free to ping me here or on gitter if there are any specifics I can help with.

[tkhyn]

@bryanrsmith aurelia-router's tests use a simple History mock class which does nothing. That's the approach I had started to follow but I saw it led to fat ugly mock classes and I thought you guys would know a better way! If I have the time to push it further I'll open another issue (or submit a PR if I manage to do something clean enough).

Regarding the issue, I'll leave you amend the code when it suits. For the moment I'm quite happy with hashChange = false!

[EisenbergEffect]

Closing this since it's a duplicate.