A UX question about requiring users to pass in --classic

Sergio Schvezov sergio.schvezov at canonical.com
Fri Jan 27 18:48:14 UTC 2017

On Fri, 27 Jan 2017 17:40:13 +0000, Adam Stokes wrote:
> Since releasing the conjure-up snap I have gotten a few questions as to why
> we have to pass in --classic when the snapcraft.yaml defines the
> confinement mode already.
> I understand that this is similar to if a user was to snap install a snap
> that was strictly devmode. We do want make the user aware of what they are
> installing and any possible caveats that go along with that. Forcing the
> use of --classic and --devmode make sense in the overall picture, however,
> cosmetically and user happiness (i guess?) this just seems like a _lot_ of
> typing.
> So I'm not arguing the use of --classic or --devmode but what if we take
> another approach and treat both --classic and --devmode as a 'force/yes' in
> the apt world and provide a simple Y/n prompt asking the user if they are
> sure they wish to install said snap because of it's current confinement
> mode?
> I much rather advertise running:
> $ snap install conjure-up

This *should* be what you advertise and I think snapd 2.22 solves that.

> And the experience be:
> Fetching info..checking confinement mode..
> This is a classic snap, are you sure you wish to continue? [Y/n]
> conjure-up installed
> Thoughts?

+1 (if on an interactive shell), if not it should fail with a nice message telling me this needs explicit concent as it is a classic confined snap.

Sent using Dekko from my Ubuntu device

More information about the Snapcraft mailing list