Roadmap 2019

[Norman Feske]

Hi Cedric,

> Anyway here's the micro-blog as it is now:
> 
>   https://tts-genode.neocities.org/genode.html
> 
> Will migrate the entries in there to genodians.org when is ready -- looking forward to it yes. More entries are queued up to help 
> "prime the pump" if need be.

I'm happy that you are on board with genodians.org. :-) Thanks for the
link to your blog, which is good read. Let me briefly comment on a few
points you raised there.

Regarding your question of native apps vs. ported apps, I share your
sentiment of preferring native apps, mostly because they tend to have a
smaller footprint and follow the design philosophy of the underlying
system. However, at the current stage, we haven't yet found the "Genode
way" of GUI applications. It is great that you support my proposed
experimentation phase. During this phase we can explore the pros and
cons of multiple approaches. I see three approaches in particular.

First, the reuse of the Haiku API is intriguing to me at the surface
level (I admittedly don't really know how it looks like under the hood).
I like the clean and consistent design language, the efficiency, and the
focus on ergonomics.

Second, the use of Qt bears the biggest potential regarding the reuse of
existing software. E.g., I'm thinking of the port of Libreoffice to
Haiku, which would certainly be unrealistic without Haiku's Qt support.
So in this line of thinking, Qt is certainly the quickest path to a
practical desktop environment for Genode, accommodating scenarios like
the office clerk described by Guido.

Third, I'm quite eager to further explore the potential of
multi-component GUI applications where the (complex and fragile) widget
rendering is sandboxed and completely isolated from the application
logic. The underlying design principle is currently employed be Genode's
custom window manager and the menu-view-based applications (i.e., the
Leitzentrale of Sculpt). Right now, the implementation of such
applications is quite unusual (if not to say awkward). But maybe the
"librarization" of init as requested by Ben would make this approach
feasible for a broader use? Now that I'm writing this paragraph, I feel
that this topic deserves some coverage at genodians.org ;-)

Maybe we even find that a combination of these approaches is beneficial?

Another area of experimentation is the secure exchange of date between
applications (copy-and-paste, drag-and-drop).

I'm very curious towards your planned line of work. The prospect of
bringing both projects Haiku and Genode together is very cool!

Regarding your remarks about acpica and usb-hid, please don't hesitate
to open issues at GitHub so we can discuss the problem (and solution) there.

> Considering the conflicting imperatives -- prioritizing security on the one hand, and some quote-unquote "urgency" to enable 
> developers to  "show off" on the other hand... Maybe it would make sense to prioritize implementing (a rough version of) the 
> feature, but keep it hidden behind a compile-time flag? There would be a prominent comment near the if-def, "leave this 
> commented out in production!"
> Just a thought. I can use a digital camera to shoot the screen in the meantime.

As interim solution, I'm using Intel AMT to take screenshots (one can
use AMT to connect to another machine via VNC). For obtaining pictures
of individual GUI features, I often start a scenario on base-linux (or
running Genode in Qemu) and take a screenshot under Linux.

Cheers
Norman


-- 
Dr.-Ing. Norman Feske
Genode Labs

https://www.genode-labs.com · https://genode.org

Genode Labs GmbH · Amtsgericht Dresden · HRB 28424 · Sitz Dresden
Geschäftsführer: Dr.-Ing. Norman Feske, Christian Helmuth