RFC: Snappy shell

Gustavo Niemeyer gustavo.niemeyer at canonical.com
Tue Jun 16 13:51:56 UTC 2015


On Tue, Jun 16, 2015 at 5:44 AM, Loïc Minier <loic.minier at ubuntu.com> wrote:

> You're suggesting to kill the idea entirely; that might be a valid
> standpoint, but you've dismissed this a bit quickly.
>
> First, consider that the networking industry is proposing this today: in
> all products, for all major brands, you end up in a custom CLI experience.
> I certainly agree it's painful to learn a new one each time, much like it's
> a pain to learn how to use/configure a new piece of software of any kind.
>

Right, exactly.


> But at least we get to define this one and offer it as a base for others
> to derive from. Perhaps frameworks/snaps could extent the command set with
> additional commands, e.g. to manage the ASIC, or provide hardware
> diagnostics etc.
>

Can we define something that is less painful instead of assuming it has to
be painful?

Second, you've dismissed the other benefits towards delivering a more
> locked down user experience (e.g. I want to ship a critical piece of
> hardware based on snappy, they may/may not install apps, they may/may not
> change the config of snaps, they may/may not run random shell commands).
>

Repeating what I actually proposed: So, here is an idea: rather than
redoing the whole shell, can we identify what are the winning aspects of
that integrated shell (good help? pleasant command names? etc) and try to
replicate that within a traditional shell?


gustavo @ http://niemeyer.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/snappy-devel/attachments/20150616/671c2211/attachment.html>


More information about the snappy-devel mailing list