Re: GNOME Online Accounts 3.34 won't have documents support



On Wed, 2019-01-23 at 10:27 -0500, Matthias Clasen wrote:


On Wed, Jan 23, 2019 at 10:03 AM Bastien Nocera <hadess hadess net>
wrote:
On Wed, 2019-01-23 at 14:33 +0000, Allan Day wrote:
Bastien Nocera <hadess hadess net> wrote:
<snip>
Flip it on its head and please suggest why, nowadays, any
application
developer, whether for a GNOME application or a third-party,
would
spend time integrating services into gnome-online-accounts, or
using
gnome-online-accounts for functionality that's somewhat core to
the
application experience, when the rug can be pulled from under
your
app
at any point?

That's not what's happening here. Until very recently, Debarshi
was
the Documents maintainer, and he's obviously been fully involved.

It is what is happening in GNOME Online Accounts in general. Pocket
is
disabled in Fedora 29, and there's a good chance that the mail
configuration bits will be disabled in Fedora 30.

I don't know whether those changes will also be done upstream, but
the
result will be the same, it won't be possible for applications
shipped
through Flatpak to know that certain configuration options will be
available in GNOME Online Accounts.


I believe in the larger picture, this is a logical consequence of
taking the boundary between desktop and apps seriously.

Right, and a functional regression for applications (core or not) that
relied on GNOME Online Accounts to enumerate services the user had
setup and authenticate with them.

As Emmanuele mentioned, the problem isn't so much that services will
disappear from under the applications (but it's a problem nonetheless),
it's that there was no communication explaining that applications
shouldn't have relied on GNOME Online Accounts in the first place, as
the functionality could disappear for reasons not caused by those
services, or applications.

It is just not right to give all 3rd party apps that show up in a
flatpak access to the GNOME api keys and identity.
They need to use their own keys. Offering a centralized service for
storing such keys, as Emmanuele suggests,
might still be useful. 

With a per-app key usage, you end up losing the single sign-on, for
example if you had 2 separate contacts applications, you'd need to
login twice to authorise 2 contacts applications separately.

The only thing you'd gain from this is that the application wouldn't
need to reimplement the authentication. It would greatly reduce GNOME
Online Account's "system" integration, turning it into a library to
help with authentication.

My own take away from this thread is that, as an application developer,
I shouldn't count on GNOME Online Accounts at all for my app's core
experience. Is that a fair assessment?



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]