"As sad by Ttiago Vignatti on his blog,"
Phoronix: MeeGo-Harmattan Is Handling FreeDesktop.org X
While Intel is looking to use Wayland on MeeGo Touch for their mobile/embedded purposes, the Nokia side is still focusing upon X for the time being. But rather than using X with KDrive, developing all of their X support out of the mainline trees, or going down any other messy paths, they are working towards using the mainline X.Org Server as found on FreeDesktop.org along with the other X libraries...
"As sad by Ttiago Vignatti on his blog,"
How do they plan to handle the most common problems with X? For example, bad flickering / corruption during transition into the X server. To get a perfectly seamless boot, they would have to implement KMS. To implement KMS, they have to export GPL symbols. To export GPL symbols, they have to actually release their kernel-mode driver as open source, and (according to some kernel hackers) are then also obligated to release their user-space code as well.
So this could mean that whatever hardware platforms they're targeting will get a nice production-quality KMS driver? Or am I just being too hopeful here?
I don't think such a small purely cosmetic glitch is a reason to create a weird X fork.
The flicker may be infrequent and only a minor glitch, but users will complain, and think it's a bug or an indication of unprofessional software design. You simply don't leave this kind of thing alone in a professional product.
Windows Mobile phones don't flicker. iOS phones don't flicker. Android phones don't flicker. Symbian phones don't flicker. Tell me, how many users will be able to understand that the reason you can't eliminate the flicker is that there was a design decision to use UMS, leading to a design limitation that can't be overcome without rewriting your graphics stack? Of the subset of those users who know what you're talking about, how many will find it acceptable that you took the easy path rather than doing things right?
Largely thanks to Apple, many users have set it in their mind that every frame of the smartphone experience needs to be precisely controlled to be correct, fast, and "whizzy". We set this requirement for smartphones, I think, even above and beyond what we set our expectations as for desktop computers. Sure, not everyone cares, but those many who do are a significant portion of the market, and they will be returning their MeeGo-based phones as defective if it flickers nastily and introduces random corruption on the screen for several seconds (during the mode switch) on boot.
MeeGo can still opt-out of trying to solve the seamless boot problem, but due to this and other limitations on X (such as the protocol overhead increasing latency), Android didn't use X. In fact, every Linux-based smartphone project I've ever heard of didn't use X due to its limitations, its size, and its general unsuitability for a smartphone. Couple that with the legal requirement to release an open source graphics driver if you want to implement KMS, and most people say "ah, screw it".
The only project I know of that actually tried to use X.Org Xserver on their phone, was ACCESS Linux Platform, and they never shipped their software on a device. They might still be trying, but AFAIK they don't have any device manufacturer customers at the moment. Maybe Nokia and Intel can do it better? I don't think the problem is with the companies trying to use it; the problem lies within the design of Xorg itself.
The N800 and N810 avoided the boot flicker via a tiny hack to preserve the framebuffer contents while booting, which I believe is still present in the N900.
Also, every Maemo device has had a fully open source driver for the display engine (drivers/video/omap for 770-N810, drivers/video/omap2 for N900) merged into the mainline kernel already. It's not (yet) KMS, but still.
Yes they do. On some phones it's to white and back, on some the backlight briefly flickers on boot.Symbian phones don't flicker.
I can't speak for Android or iOS, but I'd bet they don't have perfect boots either.