-
@3arn0wl Also watch today´s Q&A we might have some answer for you there.
-
@Flohack - It was great to see UT working on the Pine board in yesterday's Q&A. From what Jan has said recently about Google's direction with Fuchsia, the development of a completely Open alternative becomes vital.
-
@3arn0wl Yes I cannot repeat it too often: Android Fuchsia threatens the last part of openess that Android currently has: By removing the Linux kernel they are free to go fully closed source. Even if their microkernel might be open, all high level functions can then nicely be wrapped in closed source modules.
The idea of a microkernel seems neat for phones, true: Lots of features of a fully blown Linux kernel might never be used on phones at all. But its more about the political decision here than a technical one. So yes, UBports is fully aligned with fiinding alternatives, seeing Android porting as a temporary measure as the underlying problem cannot be fixed at all.
BR
-
Even if their microkernel might be open, all high level functions can then nicely be wrapped in closed source modules.
So no different than currently. All the high level stuff in Android is closed. It's why we can't ship the foundation services to be able to hook up push notifications, which are already implemented in most web services for Android, and thus must have our own implementation which almost nobody supports. Beyond that, it will basically not be any different than needing something like libhybris, but for Fuchsia, instead.
-
Beyond that, it will basically not be any different than needing something like libhybris, but for Fuchsia, instead.
The permissive license has potentially vast implications.
-
@trainailleur said in PinePhone:
The permissive license has potentially vast implications.
Feel free to elaborate. The drivers aren't going to be any less free, and that's where the real issues are. We've already got overly proprietary drivers with a GPLv2 kernel, so I don't see how having the kernel be Apache 2 is going to make it any worse, considering vendors already don't respect the GPL in most cases. All it does is put Google themselves in the position of not having to worry about the kernel license versus proprietary drivers.
Beyond that, nothing else is really going to change. Most of the middleware that we do need to run in the android container for UT is already Apache 2 anyway, and that's unlikely to change with Fuchsia underneath.
Really, all I see is people being hysterical for no reason.
-
The drivers aren't going to be any less free, and that's where the real issues are.
I guess that's why it's important that the Pine and Librem efforts are successful: to demonstrate to hardware manufacturers that there's a strong desire for open devices, to encourage them to produce more.
There are efforts though to produce Open Source drivers, aren't there?
-
@3arn0wl said in PinePhone:
There are efforts though to produce Open Source drivers, aren't there?
Yes, of course, and for the A64 such drivers are quite far along too [video showing open source hw video acceleration] : https://youtu.be/X18JN1pq2H4?t=1
Then there is lima for the GPU (WIP): http://linux-sunxi.org/MaliNot saying that initial releases will not use blobs - they probably will - but in time devs will surely switch over to FOSS drivers.
-
Well that's very encouraging news, @PINE64 -
@3arn0wl said in PinePhone:
There are efforts though to produce Open Source drivers, aren't there?
For Pine64 systems, yes. But Pine is not based on Fuchsia. The previous comments were about how Fuchsia would be "bad" for things like UT. I don't think it will be any worse for alternatives to Android, than it is now, and was trying to explain that.
There are also efforts to reverse engineer drivers to create open source ones in upstream kernel, for existing Android devices too. But it's extremely tedious to do. Unless the hardware manufacturers themselves are working to provide Open Source drivers, like Pine64 are, it's not going to matter much though.
Point is, let's stop worrying about Fuchsia and other future possible Google Android things, and just work on making UT be the best it can be, on whatever hardware/kernel/drivers we must use.
-
@trainailleur said in PinePhone:
The permissive license has potentially vast implications.
Feel free to elaborate.
First, note I used the word "potentially." Then, if you disagree, perhaps you could elaborate on how the implications of a permissive license cannot possibly be vast.
The drivers aren't going to be any less free
This may be so, or it may not be. We simply don't know yet. More and more embedded hardware is seeing mainlined drivers. If this is only because Linux (deliberately) lacks a stable target for drivers and OEMs are finding it easier to live with the accelerated pace of kernel development if they mainline (which seems to be Jon Corbet's supposition), the situation may be very different with Zircon.
Or perhaps it won't be, but I don't see how anyone has the information now in order to say. Even Google cannot know at this point what wrestling matches they may face with OEMs in the future.
We've already got overly proprietary drivers with a GPLv2 kernel, so I don't see how having the kernel be Apache 2 is going to make it any worse, considering vendors already don't respect the GPL in most cases.
For the moment, Google does appear to be dragging the OEMs kicking and screaming towards modern kernels, so if Google sticks with Android, we have reasons to believe the situation with Android drivers will improve as driver maintenance becomes more critical.
Beyond that, nothing else is really going to change. Most of the middleware that we do need to run in the android container for UT is already Apache 2 anyway, and that's unlikely to change with Fuchsia underneath.
Are you thinking to run the Linux kernel in a hypervisor on top of Zircon, or are you thinking of porting UT to Zircon (or something else that's not occurring to me)?
Really, all I see is people being hysterical for no reason.
Which is why I'm neither hysterical about it nor posted hysteria. But given how it's been eleven years since Android debuted on the market and we have only a handful of Android alternatives which will run on Android hardware, and only on a handful of devices, I won't blame anyone who does feel hysteria over the uncertainty surrounding a potential sea change in the OS and readily available hardware that will be available in the future.
Edit to add (since I didn't refresh the thread while composing and just saw this):
Point is, let's stop worrying about Fuchsia and other future possible Google Android things, and just work on making UT be the best it can be, on whatever hardware/kernel/drivers we must use.
I agree completely
-
@dobey The difference is that by now we got a fair knowledge to fix and bandaid stuff that breaks or is not working because the Linux kernel is something that many developers came in touch with.
Fuchsia will be all-new so we dont find forks, patches and other stuff on the planet. I assume many hundred man-hours will be necessary to bring a Fuchsia port to what we have now with core devices. A huge waste of our time.
I would say, better not do it at all.My languages: 🇦🇹 🇩🇪 🇬🇧 🇺🇸
-
@trainailleur said in PinePhone:
First, note I used the word "potentially." Then, if you disagree, perhaps you could elaborate on how the implications of a permissive license cannot possibly be vast.
I didn't say there cannot possibly be implications. Nor did I speak to any vastness of them. I was merely pointing out that it's irrelevant and not something we should worry about. Nobody is using Fuchsia yet. And we don't support 100% (or even 10%) of existing Android devices. So the argument of we won't be able to support these unknown devices is pointless. And also, I was pointing out that the issues we do have to deal with are 99% of the time, above the kernel level, and Android manufacturers already fail to comply with GPLv2 by not releasing kernel source, and by not developing free drivers, or even free modules that load proprietary blobs (like NVidia and others do).
We can worry about Fuchsia when we have support for UT on as many devices as LineageOS supports, and there are actually shipping devices we would like to have UT support on. Until then, any mention of it is a waste of time.
Are you thinking to run the Linux kernel in a hypervisor on top of Zircon, or are you thinking of porting UT to Zircon (or something else that's not occurring to me)?
The only Zircon I know of is an IRC client in tcl/tk with a shared whiteboard feature. I presume that is not what you're talking about. I'm not sure what you're trying to infer here. I'm not the one who brought up Fuchsia in this thread. I was simply stating it's not something we should be worrying about.
Which is why I'm neither hysterical about it nor posted hysteria
IMO, any post speculating on the woes of Fuchsia relating to UT is hysteria, because it's all speculative and doesn't help make UT better, considering there are millions of existing Android devices which don't have anything to do with Fuchsia, and we only support about 10 of them.
-
I assume many hundred man-hours will be necessary to bring a Fuchsia port to what we have now with core devices. A huge waste of our time.
I would say, better not do it at all.So why does it keep coming up in discussions? We aren't talking about porting UT (or unity8) to Windows, OSX, iOS, Windows Mobile, or any other theoretical new mobile operating systems which do not exist yet, so why do people keep talking about Fuchsia as if it's an important thing we must concern ourselves with right now?
-
@trainailleur said in PinePhone:
First, note I used the word "potentially." Then, if you disagree, perhaps you could elaborate on how the implications of a permissive license cannot possibly be vast.
I didn't say there cannot possibly be implications. Nor did I speak to any vastness of them. I was merely pointing out that it's irrelevant and not something we should worry about.
Okay. It wasn't clear to me from your three paragraph reply to my single sentence that you were making a different point instead of a direct reply to what I said.
Nobody is using Fuchsia yet. And we don't support 100% (or even 10%) of existing Android devices. So the argument of we won't be able to support these unknown devices is pointless.
Not an argument I made.
And also, I was pointing out that the issues we do have to deal with are 99% of the time, above the kernel level, and Android manufacturers already fail to comply with GPLv2 by not releasing kernel source, and by not developing free drivers, or even free modules that load proprietary blobs (like NVidia and others do).
Not something I disagree with, though as noted earlier, we have reasons to believe the situation with Android devices will probably improve, so long as Google maintains its interest in Android.
Until then, any mention of it is a waste of time.
I don''t see that curiosity and speculation about the plans of the dominant force driving the non-Iphone mobile hardware market is a waste of time. Also, this thread is in the off-topic section.
Are you thinking to run the Linux kernel in a hypervisor on top of Zircon, or are you thinking of porting UT to Zircon (or something else that's not occurring to me)?
The only Zircon I know of is an IRC client in tcl/tk with a shared whiteboard feature. I presume that is not what you're talking about. I'm not sure what you're trying to infer here.
Zircon is the kernel used by Fuchsia. When you indicated that little would change in the realm of porting to new devices under Fuchsia, I was curious on your thoughts of how porting would change with a new kernel. Given what you've said since on the topic as a whole, I gather you're probably not interested in elaboratiing. I understand, but I would nonetheless find your thoughts interesting if you ever wanted to share them, as you know rather a lot about porting and the guts of the existing UT OS.
-
@trainailleur said in PinePhone:
so long as Google maintains its interest in Android.
I don't think they're working on an alternative kernel and stack, in order to maintain interest in Android, going forward. Granted, it might still be called "Android" when they switch, but an incompatible new version.
Also, this thread is in the off-topic section.
Also, this thread is titled PinePhone, regardless of what section it is in. Nothing to do with Fuchsia, so why was it brought up here? Seems like discussion about that should be in a thread about it, not a thread about a platform meant to be using upstream Linux kernel, rather than Android or Fuchsia.
I was curious on your thoughts of how porting would change with a new kernel.
Basically, porting to any device at the moment is about the same as "porting to a new kernel." All Android devices use different kernels, with different configurations. Once all the low level software which interfaces with any new kernel is ported to work with that kernel (or in this case, new OS), then it's simply a matter of having the software built to work with that kernel (OS). Likely, we will need to have a Mesa-only (Linux upstream) rootfs and an Android rootfs, for the devices we're currently working to support. If we wanted to support Fuchsia or Windows or anything else, we'd need a different rootfs to work on top of those operating systems instead.
-
@trainailleur said in PinePhone:
so long as Google maintains its interest in Android.
I don't think they're working on an alternative kernel and stack, in order to maintain interest in Android, going forward. Granted, it might still be called "Android" when they switch, but an incompatible new version.
Yeah, that's what I'm thinking too.
Also, this thread is in the off-topic section.
Also, this thread is titled PinePhone, regardless of what section it is in. Nothing to do with Fuchsia, so why was it brought up here? Seems like discussion about that should be in a thread about it, not a thread about a platform meant to be using upstream Linux kernel, rather than Android or Fuchsia.
Yeah, it is an odd thread drift. Maybe a moderator could split the thread, leave the Fuchsia stuff here and move PinePhone topic out off off-topic? Not sure why it's here instead of in Porting, General, or News.
Anyway, I'll make this my last Fuchsia post in the PinePhone thread (unless/until in the future PinePhone supports Fuchsia.
)I was curious on your thoughts of how porting would change with a new kernel.
Basically, porting to any device at the moment is about the same as "porting to a new kernel." All Android devices use different kernels, with different configurations. Once all the low level software which interfaces with any new kernel is ported to work with that kernel (or in this case, new OS), then it's simply a matter of having the software built to work with that kernel (OS). Likely, we will need to have a Mesa-only (Linux upstream) rootfs and an Android rootfs, for the devices we're currently working to support. If we wanted to support Fuchsia or Windows or anything else, we'd need a different rootfs to work on top of those operating systems instead.
Thanks for elaborating. Good to know your thoughts on this.
-
This topic has now off topiced it's self so it will be locked. I would very politely suggest that as and when more info or developments concerning UBports, Ubuntu Touch and PinePhone surface a new thread is created in the correct Forum section. Thank you for your understanding.
