On Tue, Jul 7, 2015 at 6:19 AM, Benjamin Francis <[email protected]> wrote: > On 7 July 2015 at 07:12, Jonas Sicking <[email protected]> wrote: >> >> So in practice, if we don't do any of this, and simply treat pins as >> bookmarks, there will be relatively few downsides. >> >> If, on the other hand, we stick pins in the application registry, the >> main thing that would immediately happen is that pinned content would >> get its own cookie jar. I.e. as soon as you pin something you'd get >> automatically logged out. >> >> We could definitely change that. But given the relatively few >> downsides of just using the bookmarks database I think we should do >> that for now. That way we'll have more freedom to modify what is >> stored since it can be handled entirely within Gaia. > > > I think I agree... we could deliver an MVP for pinning sites which provide a > W3C manifest (or no manifest) without needing to use the Apps API or a new > Gecko API, it's feasible to handle them entirely within Gaia. > > But how should we handle the case where the user navigates to a Firefox App > in the browser (either a legacy hosted Open Web App or a new signed > streamable packaged app), and then tries to pin it? Would we a) just pin it > like a web site using metadata provided by the page and not provide any > additional features, keeping the existing separate install flow in the > Marketplace or b) require that all Firefox Apps provide a manifest link > relation in their pages pointing towards a manifest with proprietary Mozilla > properties, somehow detect the difference, then internally install the app > via the exiting Apps API? > I would have a strong preference for b), but I'm not sure how feasible that > is from a BizDev/Evangelism point of view. We'd need to be clear on how this > is going to work very soon to set those balls rolling.
I don't think it's at all feasible. And it would still cause the user to get logged out when you pin the site, which likely means that it's worse for the user than simply doing a). I don't think very many hosted apps use the extra permissions which means that the difference between a) and b) is very little, other than that b) logs you out. For streamable packaged apps in 2.5 we're most likely going to grant apps all the needed permissions upon simply navigating to the app. So again b) makes very little difference other than logging the user out. >> > We might say that the biggest difference is that a site no longer >> > requires a >> > manifest in order to be pinned/installed. But all of the features we've >> > talked about which require Gecko to know about a pinned site actually >> > require a manifest anyway. >> >> I don't think that's true. We can get a nice-looking icon and a icon >> name using just <meta> and <title> tags. I would expect that that >> covers by far most of what people will put in the manifest. >> >> And there are *far* more websites out there that has that has >> apple-touch-icon <meta> information, than have manifests. > > > I think you may have misunderstood me a little here, I was saying that we > shouldn't require a manifest in order to be pinned, but we will use any > metadata provided (including a manifest), Ok, cool, sounds like we're on the same page. > and it's mostly Firefox Apps that > Gecko needs to be made aware of. I'm not sure what you mean by this. I definitely think that we should make use of the apple meta tags to the same extent that we do manifests. I believe the fennec team has found that the get a much better user experience for their "bookmark to homepage" when making use of the apple tags. / Jonas _______________________________________________ dev-b2g mailing list [email protected] https://lists.mozilla.org/listinfo/dev-b2g
