ui-sref, html5mode, angular 1.4.4, IE9 issue

[RobbieTheWagner]

When I use ui-sref and html5mode, it does not correctly add in the hash, so the href is /join instead of /#/join.

It does appear to "try" to redirect to the correct page for the state, as my join page flashes up for a second, but it never actually changes the url, so it appears to redirect back to the root /.

Help?

[nateabele]

When I use ui-sref and html5mode, it does not correctly add in the hash

.

😉

Do you mean when you turn HTML5-mode off?

[RobbieTheWagner]

@nateabele no I meant what I said.

I set html5Mode(true). I'm talking about the fallback for IE9. It doesn't work.

[nateabele]

Oh, wow, I completely forgot that was a thing. Honestly I'm really not sure. Most of the hashbang/HTML5 logic is handled internally by $location.

[RobbieTheWagner]

Well all I know if our original app using angular 1.2.28 and the normal ngRoute worked as expected. If ui-sref correctly calculated the href with the # in it, it would work.

[RobbieTheWagner]

I just confirmed it's still an issue when using Angular 1.2.28. isHtml5 seems to be logging as true for IE9, and it should not be, as it is supposed to fall back, right?

[nateabele]

isHtml5 is set based on the value of $locationProvider.html5Mode(), which seems like it only returns whether you've enabled HTML5-mode in your app, not actually whether or not it's active.

Unfortunately it doesn't look like the $location service exposes anything to even tell you whether you're in fallback mode, and worse yet, it appears to violate CQRS: you can't ask it to generate a URL, you can only ask it to update the URL in the browser.

I'm kind of at a loss as to how to move this forward. We might have to kick it up to the Angular team.

@christopherthielen @eddiemonge @wesleycho Can you guys think of anything I'm not thinking of here?

[wesleycho]

Yeah, this probably requires a hook from $locationProvider - maybe having Angular expose a private variable like $$html5 that is set to a boolean depending on the browser & html5Mode?

[RobbieTheWagner]

Could perhaps do a check for window.history.pushState?

[nateabele]

That's a pretty big worst-practice in Angular land since it's untestable.

[wesleycho]

Not only that, but it is sure to cause other defects if we were to monkeypatch that method since it is guaranteed to cause unexpected behavior.

[RobbieTheWagner]

It's a better solution than using the wrong hrefs though

[wesleycho]

No it isn't, it would cause huge regressions.

[RobbieTheWagner]

I don't see this being the huge deal you guys are making it out to be, but I'll take your word for it.

[nateabele]

I guess that's a matter of perspectives: you have to deal with your use cases; we have to deal with everyone's. And, frankly, this is the first time this has ever come up.

[RobbieTheWagner]

Well most people probably don't care about IE8 or IE9 or, for that matter, IE in general.