ible
professional vim user
or a fast-motion video of ED assembling the first 100. see him start out slow, then get faster...
then heavy breathing gets faster...
then heavy breathing gets faster...
This. I really enjoyed the first assembly vid and the case build seems like a walk in the park.By the way he recorded video with one hand and messed with the pyra parts with the other hand on the last video...I take assembling it should be lightning fast and easy.
I have started to hack something:So, I'm looking forward to someone creating a driver for this.
If I have a MicroSD to boot Android from, will this interfere with that?I have started to hack something:
http://git.goldelico.com/?p=gta04-k...b;hb=48068d5e712157490ca0647922ead9e90e20677f
Currently it is organised as an SDIO client driver (like for the WL1837 WiFi chip). But that seems not to be the right way to properly intercept all mmc block read/write commands. And it tries to match the client device's vendor id while the switch is transparent.
As long as such a driver is included in your Android kernel, it will work. And it must be included, since everyone who wants to provide an Android image (e.g. Replicant) for the Pyra will have to use the Pyra kernel... Not only for this switch but for a lot of other drivers: charger, display, keyboard, ...If I have a MicroSD to boot Android from, will this interfere with that?
If I was using Android I wouldn't want it to be able to see the internal storage, it might try to write to it.As long as such a driver is included in your Android kernel, it will work. And it must be included, since everyone who wants to provide an Android image (e.g. Replicant) for the Pyra will have to use the Pyra kernel... Not only for this switch but for a lot of other drivers: charger, display, keyboard, ...
It only would make trouble if you use a system that does not use the Linux kernel (e.g. FreeBSD).
I have started to hack something:
http://git.goldelico.com/?p=gta04-k...b;hb=48068d5e712157490ca0647922ead9e90e20677f
Currently it is organised as an SDIO client driver (like for the WL1837 WiFi chip). But that seems not to be the right way to properly intercept all mmc block read/write commands. And it tries to match the client device's vendor id while the switch is transparent.
As long as such a driver is included in your Android kernel, it will work. And it must be included, since everyone who wants to provide an Android image (e.g. Replicant) for the Pyra will have to use the Pyra kernel... Not only for this switch but for a lot of other drivers: charger, display, keyboard, ...
It only would make trouble if you use a system that does not use the Linux kernel (e.g. FreeBSD).
i though we would all be using IOS on the Pyra so apple can tell us what we can and cannot haveIf I was using Android I wouldn't want it to be able to see the internal storage, it might try to write to it.
Before getting anything mainline it has to be developed... And the mainlining process can take years, even if your development "just works". So it will not be in mainline first.I thought the idea was to have most of the stuff mainline. Has this changed?
Yes or no. Depends. If MLO (= SPL = Second Phase Loader) comes from eMMC it still can throw the switch to look for U-Boot on both. This is a similar driver as for Linux but much simpler. Doesn't need to provide both in parallel. But also has to be developed.That sounds like only one or the other will be available at boot.
Oooo, I would run my Pyra at top speed just to get that warm-wood-filament smell all the timeYeah, why waiting for a case when there is a 3D printer next to me that wants to print one with a wood filament
If I understand correctly, the switch acts as a ganged relay. When power is not applied, the eMMC is connected. When power is supplied to the switch, the uSD is connected. In a power failure scenario, it should default back to the eMMC.I wonder if there's a problem with unplanned power outages. Those should be less common on the Pyra than on the Pandora due to its battery measurements actually being incorporated into the design, but I've still had occasions where the only reasonable way to get out of a badly crashed fullscreen process that was hogging the CPU was to yank the battery on purpose. If it was multiplexed to the SD card at the moment power goes down, but that doesn't have a bootloader on it.
I guess that depends on how long the power's out for though. If the system goes cold, then the switch will have reset to pointing at the eMMC I assume, so if it doesn't come back straight away, just pull the power and leave it for a bit longer before putting it back on.