XO-tablet development? [Devel Digest, Vol 90, Issue 8]

Yioryos Asprobounitis mavrothal at yahoo.com
Wed Aug 14 04:52:33 EDT 2013


> By proposing that further development work be native to Android,
> you are locking the fruits of that labor away from:
> - Any child using a current XO laptop
> - Any child using any other Linux laptop, such as the millions of children in 
> Brazil, Venezuela, Argentina, etc...
> - Any child using a Windows laptop

As far as I understand the question was which way the OLPC laptop/tablet to proceed in the future.

The issue then would be is it likely to reach more people with the current +HTML5 course or with Android?

Regarding benefits to the existing XOs, ARM-based should have no problem. XO-1s can hardly benefit from any future development both because of age and also performance (13.2.0 is already hardly usable on these). So is only the (how many?) XO-1.5s we are talking about.
Regarding the millions of other x86 machines, I would argue that only the power savings from ARM machines would be sufficient to pay for any cost to replacing them ARM machines. Looking at the future and the continuously increasing power needs for computing, I would say is almost mandatory if you want any kind of saturation.

> Why would you do that ?     By working within the proposed framework (Sugar on 
> HTML5),
> these children are supported as well as those using a Android tablet.
> As James has already pointed out, the performance penalty of using
> HTML5 is minimal --- probably less than that of using Python on most systems,
> as much work (independent of OLPC) goes into optimizing its performance.

I do not know if HTML5 can regenerate the Sugar stuck and vital aspects of it as collaboration and the journal. Can it?
What about some more complex activities like turtleart, or 3rd party apps like etoys, wordprocessing (without server-side support) etc?
If we are  really talking a small number of simple standalone sugar activities I would hardly see the benefits.

Regarding ubiquitousness of HTM5, is certainly better but there are at least a dozen browsers in existence, each available in different versions and each supporting different levels of HTML so you should either lock it to one browser (which one?) or keep developing for all of them or use only the least common denominator features. 
In an all HTML system for all browsers and platforms we may also need to consider the  URL security vulnerabilities that HTML come inherently with, which is in clear contradiction to the so far security scheme and the needs of the young users.

Performance wise the proposed course has another major bottleneck that  Android in contrast to Linux/GNU appears to have an advantage on. Video drivers. This is probably because Android devices are using proprietary drivers and the latest OpenGL version but trying to play HTML5 video on XO-4, is reminiscent of XO-1 with Flash. I would really like to see how it will perform with HTML5 activities to believe anything about performance.
Finally regarding HTML5 and performance/usability, the smatrphone/tablet ecosystem has a clear verdict on this I believe: Go native (Android or iOS) unless is not allowed or is a fairly rudimentary app. 
I do not know why should we dismiss the experience of several thousand developers and hundred of thousands of apps.


PS: (Not to crowd the list) 
I'm not a party animal ;) but if wasting the work was the issue then we should stick with Fedora/Gtk/Python.
Besides few months ago neither the tablet nor the unified laptop/tablet development team was in clear sight.


> Cheers,
> wad

More information about the Devel mailing list