Longevity vs Obsolescence (software)

Maybe you oversaw Samsung’s free LTE modems at replicant.us after you wrote everything from replicant is for outdate devices. Yes, they are, but maybe also because FP denied to follow replicants advice.


Also NXP/Freescales Android processor i.MX6-8 (1,2 GHz) and TI’s OMAP, and even chinese cheap Allwinner’s processors especially for Lollipop-android they all had opensource drivers. After replicant’s developer explicetly contacted FP during FP1 times to have better replicant support for FP2 I cannot understand why you neglect FP’ s reaction to replicant.

Is it maybe because you criticised only one simple citing mistake at wikipedia about replicant’s mission ?

1 Like

In my opinion, Fairphone took into advice them, but chose Qualcomm because they actually need to sell a device with a good processor, and Qcom’s fame was and is a benefit for that. If Fairphone cannot sell a bunch of devices, they won’t be able to have the manpower to update the software of their two devices.

Let’s not forget that an utopic product without market share is useless; Fairphone needs to open that space first. They are working hard on that.

And no, there’s no perfect and top chipset (powerful, fully libre and highly demanded by consumers) yet. It is hard even on computers.


This logic about ‘sell & forget’ implies that FP in the most critical issue (processor) intentionally decided against longevity and in favour to obsolescence(!). Replicant’s main developer also gave a frustrated statement about this decision: Fairness at FP shouldnot be mistaken by fairness for users e.g. in the sense of transparency about the whole device.

I am not sure if it makes sense to offer a device with an outdated processor from 2014 starting in 2016 and trying to talk around by explaining that a “flagship” processor that was already outdated was needed to get a market share.

You cannot have both as you already made clear by that valuable web link that I rather appreciated

Fairphone is clearly not forgetting any of their devices. They are upgrading the Android version of an old device (FP1) that they are not producing anymore!

That’s part of the result of spending a whole year looking for a completely-free laptop, compatible with Debian as-is (without the non-free packages) and below the 1000€ limit (Purism is just too expensive for me). Glad to share it and help! :smiley:

That’s a complex topic, there are various kinds of fairnesses! FP’s main objective is human fairness: health and working conditions; second, ecological fairness: repairable modular phone, software updates; third, software fairness: a Google-free OS, FP Open sources, community OSes. All of them are user-looking fairnesses, IMHO.
That’s my analysis, at least.


What makes it even harder on phones is that some control freaks on the other side of the Atlantic (I am European) want to be able to poke their nose into everything and have control over everything - specially concerning radio. I am only aware of one baseband processor, TI’s “Calypso” chipset that was used in some older Motorola phones that can actually be run by free and opensource software: http://osmocom.org/projects/baseband. The FCC wants to lock down access to all devices containing radio communication chips (WiFi, GSM etc.). Here are a few interesting links:


The thing (philosophy) with open source is not to sell enough for updating a proprietary by one’s own company. In opposition open source aims to give others the chance without beeing under pressure.

Therefor you don’t have to overstrain how I used ‘sell & forget’. The word forget’ was dedicated to the obsolescence of FP2 with the next android release called nougat. Yes, even you have to forget your FP2 for Nougat. If it would have been one of replicant’s cited open hardware procs you probably wouldnot have had to forget it. Every flagship proccessor producer has the opposite commercial philosphy to replace their flagship in the next selling season. After selling Qcom in FP2 FP agreed and follows straightforward the processors commercial aims. Means though they don’t forget their phones by themselves. But they can’t help to follow obsolescence rather than longviety of open source hardware.

I am aware of Samsung-RIL (actually I am using it), but which LTE modems does it support? I am not aware of any…

The NXP/Freescale is imho pretty rare and TI’s OMAP is complete legacy. Allwinner A20 is quite promising, however, this SoC does not include a modem.

Well yes, but if you have a boat with 5 holes and you fix 2 of them your boat will still sink…


OMG. We had enough Wi-Fi problems already… almost all Wi-Fi drivers are proprietary (pss, even a bunch of those included on the mainline Linux kernel). Sad world, :confounded: :sob:

Tell that to Google, the designer of Android (and the on purpose bad-named Android “Open” Source Project).
I know the philosophy, I (try to) live following it (I’m a Debian user, for goodness sake!), but there are products (and times) simply not designed for that, and Fairphone has no fault nor divine power. GNU/Linux was designed for that; Android wasn’t.
The only perfect solution for FP is to design and produce their own chipset, and that’s not possible in terms of money yet, unfortunately. They are welcoming the discussion, at least, while working on other fair matters. One step at a time.

I’ll tell you one thing related to the Replicant’s project: I respect them a lot and I recognize their labour and I’d want to own one device compatible with their OS, but CyanogenMod has done more for the mobile obsolescence with blobs than Replicant without them, in terms of devices and Android versions. Theory vs practice. (I know the labour of Replicant is incredibly more difficult that the one of CM, and they have less manpower, yes. It’s the world where we live, and the two projects are awesome on their own)

I liked your comment, and I agree with you. But, as a developer and (I think) a reasonable person (like you all), I know Fairphone cannot fix five them all. Their devices will sink, of course, but as later as possible. Let’s make their time as valuable as possible for the world.

PD: I’m searching for a Canonical worker’s paper explaining the obstacles they had to workaround while designing Ubuntu Touch using the Android’s HAL drivers for extreme compatibility. I’ll edit once I find it, it was located on a Ubuntu FTP server I don’t recall…


Well, I think it should be sufficient to just use a chipset that is as open as possible. And the current choice for FP2 is definitely not, as previous posters have indidicated here with the incredible blob-count!

Well, you can buy one of these devices in very good shape for around 30-100 euros nowadays. But don’t be disappointed once you see how they work in practice…

And since Replicant is based on CyanogenMod you probably can’t distinguish the projects this way. So Replicant benefited a lot from CyanogenMod, but Replicant also gave back the Samsung-RIL stuff (the ports in CM use this free RIL layer instead of the crappy,
proprietary and insecure Samsung RIL).

Imho, there is not so much bad in Android, and it indeed is very open. Of course, Google’s policy of not opening up the development process and just throwing some code over the wall (similar to Fairphone :stuck_out_tongue: ) has been often critized. Also, many people judge their license to be rather too open as it invites bundling the free platform with lots of non-free things. If Android had a stronger copyleft, we probably would not suffer from the situation with the proprietary blobs as we do nowadays.

1 Like

I’m not being clear: my opinion is like yours. Unfortunately, they need to sell, as I pointed above (we live in this world, they priorize to make a change in workers conditions and I can live with that and I support them with my money and time). Also note that changing from Mediatek to Qualcomm was indeed a step forward in openess, even with the huge blob-count. Just not libre enough for some other projects yet.

Oh! I didn’t know that. Thank you! Does it have something to do with this developer? As far as I recall from that interview, he worked hard on the CM’s Samsung support (mainly for userspace drivers, a.k.a. HAL components, like the RIL one)

Apache license permits not giving back to the community your work done on top of an open source software, just the thing GPL fight hard against. You could close the development of an open source software!
Also, the not-so-open in Android is just that: they designed a system of userspace drivers (called HAL components) to avoid the vendors to include their drivers in a GPL software, the kernel, Linux!
This, added to the hype Google has the ability to raise (and with what I call the “Open Source naming honeypot”), made Android the most widespread mobile OS. Apache license is a double edged weapon.

1 Like

Changing from a huge GPL-violator that is unsupportive to a company which provides blobs on a regular basis? This may be better for selling, but I don’t see any advances in terms of openness here.

Sorry, no idea.

Well from the vendors perspective it is more open as it gives them more freedom, not less. They don’t have to disclose their source code but of course nobody hinders them…

Well, but if you look at the completely locked-down alternatives from Apple, Windows or Blackberry you will have to agree that this was a really good move from Google. I doubt they would have had this success with platform that enforces more copyleft as well.

The main issue for me here is that the buyers don’t care enough about openness. Of course, FP’s main goals are fairness and longevity, but they are defeating these goals themselves by their platform choice. And the sad thing is that they didn’t discuss this choice with their community beforehand and they don’t discuss it afterwards either. If there is/was a blob-free LTE modem, imho the Allwinner A20 would have been a much better choice.

Btw.: If you follow CM14 development you will notice that they invest huge porting efforts into getting devices based on the Snapdragon801 platform (e.g. Samsung Galaxy S5) to run Android 7.0 - so I wouldn’t lose hope for the FP2 on this regard too early.

I agree. There are other projects like Novafusion for example, who try to keep older devices going. Nova for example still provides various Android releases for the Samsung S3 mini that uses the Nova-Thor chipset. So, fingers crossed, the situation is not that bad.

1 Like

That’s because of the Codeaurora Forum (CAF) sources! Qcom platform (the board) is obviously more open than Mediatek; not libre, not blob-free, but more open. You don’t see so much community ROMs for Mediatek devices.

I’m working on a CM12 port (among other great forum participants; ad: we welcome new developers!) and I’ve benefited from CAF and from other Qcom CM ports, just because I have sources to learn from.
Atilag, the man who ported FxOS to the FP2 (hi, Juan! See you on the next Mozilla Connected Devices meetup! ;)) benefited from the FP Open OS sources, based on CAF sources.

I reaffirm: you are right and they are not libre nor blob-free, though.

Of course it was. They have triumph even over latter and more open alternatives like FxOS (B2GOS nowadays) and Ubuntu Touch because of the network effect they had the foresight to take advantage of (in terms of developers and available apps).

Yes, you are right. I often call that the “fifth and selfish freedom”. They benefit from other’s work, but are (can be) unsolidary enough to do not give back their own work. Well, that’s legal and the collaborators accept that license before working on the project, but that fact makes me very sad sometimes. I’ve done that myself on a previous job and it doesn’t feel good, :frowning:

Didn’t they? :scream: Where did suggest Replicant’s members the chipset choose? I thought there was a forum topic, at least

PS: It’s a pleasure to talk with you, @kuleszdl, honestly, :slight_smile:


That’s great!

Well, not a topic that I was aware of, but if such thing exists and someone could point me to it I would be eager to read their arguments. And afaik the suggestion from the Replicant developer(s) (actually it was more or less a one-man-army back then…) came up because (potential) users of Replicant approached them and suggested to support the device. You can find a discussion about it here:


Btw.: I forgot about Paul’s response - indeed there seem to exist “libre” LTE modems.

1 Like

The following 6 posts were moved from there because they fit better here.

That’s not entirely correct.

Android wasn’t intentionally designed for planned obsolescence. The issue was that when Android was first made, carriers were unhappy with Apple’s strict policy of having an unaltered version of iOS on every Apple phone. These carriers told Google that if they wanted them to support Android, they needed to be able to do their own customization of the OS. The same demand came from phone manufacturers as well, so Google did its utmost best to make Android as customizable as possible very every party involved in producing and retailing a phone.

The end result is that there are pretty much as many versions of Android as there are different models of phones which makes it impossible for most of the companies in the chain to properly keep the OS on the devices they produce or sell up-to-date. Technically it’s possible, but it costs too much money. Apple is in a different boat, because they produce their own devices so iOS only needs to be adapted for that limited variation in hardware. Android runs on thousands of different hardware configurations.

While from the consumer’s point of view, this can be regarded as planned obsolescence, it wasn’t initially intended as such.


Hi Jerry, very interesting - thx for taking time to explain!

I disagree. Take GNU as an example: any GNU/Linux distro run in a lot more hardware than Android, and they are customizable and easily upgradable. Android is not modular and GNU is, that was the design I was talking about.

This is my point of view and I respect you disagree (that’s healthy! :slight_smile:). Obviously, your history is mostly right.

Those distros run on x86/x64 architecture though.
In the world of ARM, platform standardization is severely lacking.