Re: Maemo is CSS contaminated (not entirely open)
 

No, it's sort of a compromise right now. It's the GTK tooltip widget, same one as OMWeather uses. :\

Quim didn't really give us an answer when it came up during thoughtfix's interview (thanks again Quim and Dan!). As far as I could read, it basically came down to, "They're closed, it's not gonna change anytime soon." No real reason was given, though.

Re: Maemo is CSS contaminated (not entirely open)
 

Kinda difficult when there are undocumented hardware components. Where you have documented hardware, gnu purists already succeded in providing a better system (ubuntu, mandriva, and surely most other linux distro nowadays) than the commercial alternative (windows).

Re: Maemo is CSS contaminated (not entirely open)
 

Considering that it sucks power like there's no tomorrow and its performance is mediocre, I'd say they chose the cheapest wireless chipset ;)
(Hey, I don't blame them for that, I also bought some of the cheapest wireless adapters I could find, but they have open drivers and perform much better than the tablet's one)

Re: Maemo is CSS contaminated (not entirely open)
 

Bluetooth driver is not closed, the source is in linux kernel. What is closed is the bluetooth firmware blob but that's pretty normal (see http://www.bluez.org/download.html for various firmwares) .

Audio could be replaced relatively easily, it is possible to access audio from ARM side and linux drivers exist or is possible to adapt them. Then DSP will simply sit unused. See linux port for Palm Tungsten T/T2 (OMAP 1510) for example, I think it uses same audio codec as 770.

wi-fi is a problem, see this mail for introduction
https://garage.maemo.org/pipermail/c...er/000001.html
basically there are two closed parts with open glue (=cx3110x.ko) between. To make it kernel-independent one needs to replace umac.ko module. As I understand it the firmware blob and the umac blob fit together so using some random open wi-fi stack with the firmware may be hard (just guessing).

Re: Maemo is CSS contaminated (not entirely open)
 

Do you have a reference for that? I thought cx3110 was pretty good in that regard.

Re: Maemo is CSS contaminated (not entirely open)
 

Man, that was beautiful. Thanks.

Re: Maemo is CSS contaminated (not entirely open)
 

wgetting a big file uses 75% cpu (according to osso-statusbar-cpu), I suppose (but I may be wrong) that a sizeable part of it is due to the wifi transfer (on another arm system wget takes just 30%).
I also suppose (but again, I might be wrong) that cpu consumption translates to power consumption.
Btw, I don't really care about wget performance, I noticed the problem because it's very taxing to stream video to the tablet, in part due to missing video optimization (even with the great work done with the mplayer port) but in part due to the wifi (lack of) performance/efficiency.

Re: Maemo is CSS contaminated (not entirely open)
 

I retried with the output redirected to /dev/null (writing to the flash takes its toll on the cpu) and "Osso mcspi" takes 41% cpu, while cx3110x takes 12%. So the cx3110 isn't completely to blame (though 12% isn't that low), but overall wifi takes more than 50% cpu.

Re: Maemo is CSS contaminated (not entirely open)
 

I guess I'd be more concerned with power usage figures on the wifi itself? It uses CPU to drive it, of course, but I expect more of the power is directly consumed.

Re: Maemo is CSS contaminated (not entirely open)
 

Network transfers are CPU-expensive. This really says nothing at all about the chipset itself.