Re: [w3c/manifest] Ability to detect if the current document is part of an "installed" web app (#454)

> On 18 Apr 2016, at 10:07 PM, Adrian Hope-Bailie <notifications@github.com> wrote:
> 
> A dev can already fetch() their manifest, so what would this provide and why?
> 
> There seems to be an assumption in a lot of the spec that a user that has the manifest installed will never visit the app other than via the new icon on their homescreen.
> 
Yes. It's complicated because transition between the app experience and the tabbed browsing experience may differ - and jumping into one or the other may annoy/confuse users. We just don't know yet. It's something that will take a lot of time to test and get right. 

> My reading of @felquis comment on #417 (comment) was that he has users hitting URLs that are part of the app but have landed there directly from links on the web. He then needs to decide if he should encourage them to install the app or not but has no way of knowing if they already have.
> 

If a developer needs to encourage installation, then the browser has completely failed: progressive web apps are supposed to solve this exact problem: 

https://infrequently.org/2015/06/progressive-apps-escaping-tabs-without-losing-our-soul/

It's too early for us (browser vendors) to admit defeat here. We only have two shipping UAs and I think only Chrome is making use of installability signals. 

Firefox is just now starting to lay down the infrastructure to test installability signals. But this will take at least one or two years (our Android team is tiny, and it's only me helping on the DOM side, and I barely know what I'm doing: 🐶💻)

Having said that, the "install" event is supposed to address the "did they install the app?" question - but still give users the ability to "clear cookies/storage" to make the site forget about it.

> Two use cases/questions:
> 
> How does a dev know if the user has installed the manifest (and when) so they can try to force an update if necessary or encourage the user to perform the install?
> 
The "install" event - but again, encouraging is considered absolutely a failure from a browser vendors pov: we don't want sites annoying users with "install our app" buttons and banners. 

Re updates: As noted in the other bug, we probably need a "requestUpdate()" API. We would need to work out the details. 

> How does the dev know which manifest the user agent has interpreted as the current manifest (if any) if the user lands on a page that is inside the scope of multiple possibly previously-installed manifest?
> 
I imagine one can check with the Cache API - to see which version is download, and other cache information. However, the end-user may have modified aspects of the manifest (through the browser's provided install UI). 

I think we can only really solve the scope change problem with an update API. That way, the site can request and track precisely which version is being applied without any magical/unexpected things happen in the background. 

> —
> You are receiving this because you commented.
> Reply to this email directly or view it on GitHub
> 


---
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/w3c/manifest/issues/454#issuecomment-211374828