The huge amount of time required, but above all the fact that for every device there is some peripheral that has serious new bugs which take ages to fix.
The adaptations share a lot of parts, and you probably already knew that.
I’m not so sure the amount of work is necessarily so huge it takes too much time to get done, more like it’s quite complicated.
So i still think the argument Fewer Ports → More Time → Higher Quality is a fallacy.
In those scenarios more experience from more ports could be at least as valuable.
Maybe I haven’t been clear.
I mean that the ports are limited in number, and all on devices that share the same SoC manufacturer, the same CPU architecture (except for the shift to 64 bit), the same camera manufacturer (incidentally the same as the phone manufacturer), yet every time it takes eight months from the availability of the source code to the release of a public version.
I am a hardware developer, and I also write (now less than some years ago) code to run on that hardware. Writing a bare-metal driver for a peripheral is a matter of some days at worst. I can understand that getting the hardware to work in Linux is not quite the same thing, but if Linux stays the same, the CPU stays the same, the peripherals are mostly the same or similar, I cannot make out how can it take that long to develop the HAL to sit between the metal and the OS.
2 Likes
Regardless of the general problem - I don’t know what makes you think that the phone is no longer available in Germany? I quickly found several trustworthy sellers outside of Ebay, for example:
2 Likes
I do not think Jolla has internal knowledge of sony devices h/w to write their own drivers, instead jolla, as well as unofficial porters, use the binaries sony provides to build chstom aosp, eg sfos. Those binaries are not necessarily the same ones used in oem roms. In addition, all binaries are tailored for a specific kernel version which adds extra dependency.
Sony provides HW-level drivers. AOSP must write a layer to match these drivers to the Linux kernel and the Android layer on top of it.
I don’t understand how can it take so long, not to write this layer from scratch, but to adapt the existing layer for the 10 II to the 10 III.
Few aspects forgotten in this discussion:
-
Each “annual” adaptation is based on the current Android. 10II is using Android 10, 10III is based on Android 11. This shifts goalposts every year and porters have to work around it.
-
BLOBs are provided separately for AOSP and Sony’s own Android by QCOM, if I am not mistaken. As far as I understand, there is a partitioning of the development teams in Sony with proprietary Android and AOSP support developed separately. There is probably some communication between teams, but it is probably restricted to avoid any possible copyright issues. AOSP team is smaller and starts work on the devices when they are released already with proprietary Android.
Take part 2 as my understanding of the affairs. It maybe inaccurate or even wrong, then I hope someone knowledgeable will correct it.
But, as you can see, 1 & 2 combined lead to significant delay. Unfortunately, as discussed many times in this forum and at TJC, Sony seems to be the only manufacturer providing such legal pathway for basing alternative AOSP based ports on their devices. So far, if memory serves me right, nobody was able to list manufacturers that would allow you to downloads BLOBs from the manufacturer site and providing legal framework for flashing devices. If there are alternatives, would be good to list them. Just before making a list, I would suggest to read Developer's Intro · Open Devices to see what such BLOB distribution should match.
8 Likes
We are nevertheless talking about adapting an incremental device update to an incremental Android update. Were the team of only 1 person, eight months seems still too long.
1 Like
Usually the work of an engineer is done when its done and is not speed up by people hoping for it to be faster. Also usually engineers like to get their work done and move on to something else interesting. At leasts thats the attitude i got towards that.
i am no expert in this area so I only got two datapoints: 1) sony needs X months to build AOSP, 2) jolla needs about the same time to build a reliable port. As long as there is no competitor to jolla who does the same thing, we can just assume that this is just how long it takes. All other claims just sound very unfounded to me without data backing them up.
Also it looks like no manufactor wants to build the same device for many years, thus I do not really see obvious room for improvement.
2 Likes
1 Like
That just mentions warranty and software updates - not availability.
2 Likes
Yes, the concept of Fairphone is repairability. You can buy spare parts. This way a totally new device isn’t necessary every two years. It is not yet ideal, but certainly better than most other devices.
This discussion is about phone availability.
It is irrelevant that current users can be guaranteed to (be able to) keep their phone for 5 years if new users can’t buy it.
All things being equal, i’d probably prefer a Fairphone too (if just by a slight preference over current choices).
However, attributing it magical properties like 5-year availability helps nobody.
Well, with all their previous Fairphone models, they were available at least for two -three years. This is already a big difference with Sony, where in a year after a device launch a new Sailfisher needs to search in the aftermarket. Those who want to become a new Sailfish user always face a dilemma: to buy a yet to be supported device and put it into a drawer, or search the market for scraps of a previous gen model. This does not make overall experience better in any way.
I already made a calculation on alternative history if Jolla had restarted with FP2 rather than Xperia X. In that case Jolla would have developed ports for FP2 and for FP3 (only two ports). They’d focus their resources on just two devices, instead of four crude buggy ports (X, XA2, 10, 10 II).
1 Like
Fairphone has also had some stints of complete unavailability and ridiculously long delivery times too… so the aftermarket suddenly doesn’t seem so bad. Resellers of Fairphone are effectively nil.
But more importantly; why is it you guys keep saying time alone would fix the quality?
I consider that essentially disproven, since:
- Porting work is not going 100% full speed for each 18 month cycle; especially from after SFOS X launch to next device comes available.
- Nor does focus switch completely as a new model comes out (see release notes/changelogs - fixes keep coming).
- At the current pace, keeping up practice with porting and seeing new problems could be at lest as useful as just banging your head against it some more.
Then add to this that Fairphone has similar shitty Android-first hardware with crappy drivers.
(=potentially some unfixable issues and suboptimal quality anyway)
I’m sorry, but it’s looking a lot like time alone will not do much, if anything at all.
(Apart from having more users complain about outdated hardware perhaps…)
1 Like
and also fairphones look like sh*t
2 Likes
Well, I see here a straightforward logic: You can either disperse your resources or focus. It’s a simple trade-off, and it’s just universal. I’m pretty sure that if there were only two ports, purchasers of the first gen Sailfish X (those who bought Xperia X) wouldn’t be stuck with Android 4.4 support, because implementing it for two devices is plausible. This is a pure speculation of course, but it is grounded in logic.
Regarding the drivers, FP is a small company, and they have custom drivers for their current device. Unlike Sony devices, who would abandon driver support for their devices just in a year after launch, FP has a good record for providing their devices with newer android updates almost until the next model is available (3 - 4 years cycle). Only iPhone can compare to FP in terms of keeping their devices up to date. I would expect that a small company would be more flexible and responsive to issues, especially if there is an official collaboration. It could be that I have too high expectations. But you asked what is the reasoning behind some of our arguments, and I’m just sharing mine.
So far, the situation with Sailfish options are:
- Xperia X - obsolete Android compatibility layer;
- Xperia XA2 - controversial reports about GPS function, the function is just not reliable;
- Xperia 10 - abysmal battery life;
As people say, now choose your poison 
This only holds true if resources are too dispersed currently. And looking at porting activity, i cannot see that they would be. Sometimes banging your head against it just doesn’t help anymore.
The XA2 GPS issue turned out to be in closed firmware for example…
NB: I’m not arguing that it wouldn’t be possible to spread focus too thin; just that the current pace seems just fine (and that there could even be a backside to porting less often).
Again; these are your feelings speaking not an actual substantiated argument.
This was not done out of spite; but because the base for the port was simply too old to support the technologies needed for the Android App Support.
They would basically have re-port for the Xperia X, and users have to re-flash.
And as we seem to agree; that is not reasonable for a phone that is going unavailable.
I can’t believe people are still salty over this.
Maybe the Sailfish port for FP could/would would have been redone every year or so… but also to get some leverage on what you are suggesting it shouldn’t have to be, or there is no real upside. Mainline Linux support is the only game-changer in this area.
In a fairytale world everything is of course better. I’s not a terrible theory as such… but a theory all the same.
FTFY
0. Xperia 10 II - excellent port that wouldn’t exist if they did as you suggest.
2 Likes
Slightly O/T, but rumours suggest Sony is cancelling one of its 2022 phone range refreshes.
Fine, but what if it’s the 10 IV that has got the chop?
Just came home after 450 km car navigation today. Some glitches in wooded areas but no real problem. I think waiting for a signal staying in one spot watching GPS-Info and actually going somewhere is different.
1 Like