View Single Post
Venemo's Avatar
Posts: 1,296 | Thanked: 1,773 times | Joined on Aug 2009 @ Budapest, Hungary
#261
I'm late to the party, but I might still be interested in this work.

Originally Posted by chenliangchen View Post
1) I'm not a software guy, but will Fremantle work on a modern chipset? Qualcomm or MTK? As I understand certain components (probably majority) are close sources so how they are going to be ported? Will Fremantle or its equivalent work on a new SoC provide "reasonable" front engineering cost?
Short answer: No.

As far as I know, Fremantle isn't (and has never been) fully open. There are closed parts, problematic even on the N900, which basically prevent us from porting Fremantle to other devices, unless free / open source replacements are developed. There has been some advancements on this front (I'm not up to date), but last time I checked this was still an issue.

Then there is the issue with the (very) outdated software stack. Fremantle is built on Gtk2 (deprecated since 2010) with some extensions (hildon). Some parts of the UI are developed with a god-knows-how-old version of Clutter as well (the welcome screen, which is actually far smoother than anything else in the system), and then there is the (now also unsupported) Qt4 glue module.
And I haven't even mentioned that hildon-desktop and friends are heavily based and dependant on X11, which itself is nowadays becoming outdated in favour of Wayland.
Of course, there are parts of the UI which have never been "hildonized" ie. are leftovers from the Maemo 4 era (some settings panel IIRC and who knows what else). And then you have some built-in apps that are hardcoded to 800×480.

Even those that aren't hardcoded, are unlikely to support other resolutions, aspect ratios or high-DPI screens.

So, any effort to modernize the UI would affect basically every level of the software stack, including toolkits (replacing or modernizing gtk2+hildon, or rewriting everything with Qt), applications (making them scalable and touch-friendly) and the desktop itself (porting from X11 to wayland). And of course, creating new, high-res artwork, etc.

There have been some attempts on this:
  • IIRC there have been more than one attempts at porting hildon to Gtk3
  • Cordia (now a dead project) has envisioned a merger between hildon-desktop and the MeeGo netbook UI, but everyone lost interest when Harmattan was released
  • Even before that, there were attempts to get the hildon framework into Gnome, but I don't think they ever succeeded.
  • There was the Intel tablet UX and the meegotouchhome UX which never quite lived up to the hildon-desktop legacy, and of course neither of them were ever finished.
  • Some people attempted to rebuild a very similar UI with Qt from the ground up (there was at least a lipstick-based prototype by w00t), but again this was dropped even before I got to work on lipstick back in 2012.
  • There were LOTS of good ideas in Nemo Mobile, but people lost interest when Sailfish was released. Now it's mostly a dead project as well. I feel somewhat guilty for not being able to put in more free time into it back in 2012-2013... maybe things would have been different, but you never know.
  • etc.

I see three possible ways of moving forward:
  • Revive the work on Nemo Mobile. Get feedback from everyone in the community about how it should look & feel, etc. And then implement it using the toolkits that either already exist (lipstick, mlite, Nemo middleware, etc.) or were started but remained unfinished (glacier UI, etc.) AFAIK Nemo was never dead, but it has never reached any kind of usable state over the years. (This is not surprising, given most of its main current contributors are Jolla employees whose main priority is Sailfish and Silica.)
  • Or completely overhaul the old Fremantle software stack, bring it up to date with latest Gtk3, Clutter, etc. versions, get rid of the Nokia-specific (and never openly released) dependencies, create a nice-looking theme, and then carry on.
  • Or maybe now the Gnome stack (gnome-shell, gtk3, etc.) is sufficiently sophisticated that we can do away with all the legacy crap, trash hildon and the rest and simply run gnome-shell, maybe with a custom theme? There are lots of ideas in gnome-shell which make it very similar to hildon-desktop (in fact the gnome-shell overview basically is the same concept). Also modern gtk3 should be equipped to handle the mobile form factor (it has UI scaling and font scaling), so it's just a matter of creating touch-optimized apps.

Originally Posted by chenliangchen View Post
2) Because we are such a niche group, in order to cover design/testing/logistics/moulding/producing line/certification/transportation/warranty........ cost, the device won't be cheap. How much will you pay for a 5" landscape slider, assume it's 32+3GB, running on a Qualcomm 625 soc, and with the OS we want?
I believe 500€ would be a fair price (plus-minus 100€, depending on the quality and support of the device).

Last edited by Venemo; 2017-06-06 at 21:11.
 

The Following 15 Users Say Thank You to Venemo For This Useful Post: