Hacker News new | past | comments | ask | show | jobs | submit login
PostmarketOS Project Direction 2020 (gitlab.com/postmarketos)
88 points by ajr0 on March 11, 2020 | hide | past | favorite | 19 comments



It would be great if PostmarketOS got serious about supporting at least one recent phone completely.

The link above mentions Bluetooth and Camera as “nice to haves”. I would love to get away from google but half the point in a phone is the camera.


If the aim is to get away from Google, it is probably best to run something more Android-like, such as LineageOS or GrapheneOS. Stick to well supported hardware and it will just work. PostmarketOS is more like a generic embedded system.


How is that "getting away from google" if they're both based on Android, which is a Google-developed operating system?


Yeah, but they don't include proprietary Google apps such as Google Play services and all the crap that depends on it. Only open source stuff.


AIUI, 'Nice to have' in this case is in relation to 'technical feasibility' rather than actual useability. Cameras and modems are two notoriously closed down parts of modern smartphones, and thus it follows they are the most difficult if not downright impossible to support properly.


I was wondering, the Nvidia Shield uses a Tegra X1 chip, which if I recall correctly was nice in that it uses the full open GL rather than ES.

I remember seeing their have a Distro called Linux for Tegra:

https://developer.nvidia.com/embedded/linux-tegra

Shouldn't this make it an easy device to run something like Postmarket OS? Of course it's non free drivers. But in terms of getting a device working, I'm not sure why it isn't used more.


If one of the devs wants the device to port PostmarketOS to it, I got a spare one (its a K1 ie. the remake of the orig. Nvidia Shield Tablet. Not that they differ much).


I love this project!

What I would really love to see happen in 2020:

* Refocus and prioritize tasks according to available resources (people, time)

* Better communicate what is done and what need to be done (Roadmaps, using GitLab milestones for example) for people wanting to contribute.


Anyone using this having any recommendation of hardware?


The PinePhone is the device which receive the most attention, but I'm not sure of it's currently usable.


I have a PinePhone running PostMarketOS. Issues (remember its alpha software): A bunch of programs did not open at all such as the image viewer, web browser and some others. Camera program opened but couldn't connect to camera. No cellular radio support means no calls, mobile data, or sms. Wifi works so you can do all the fun admin and developer tasks. USB OTG also doesn't work so USB things like keyboards, mice and USB sticks plugged in via a USB C->A adapter didn't work. Sound didn't work for me But those are all solvable problems.

I haven't booted it in two weeks but I was able to get it to build and run drawterm (plan 9 terminal emulator) without much effort save for downloading a few libraries. No input as KDE Plasma uses a different input method than x11 so until a shim or patch, x11 programs probably won't work out of the box. In the case of drawterm, it connected to my CPU server and displayed the login prompt. But tapping the screen did not bring up the keyboard so I could not log in.

Conclusion: Owning a modern open smartphone is liberating. It feels like a PC in the sense you can just download distro images, write to SD, boot from said SD and even install to mmc. Once things progress you can do all of that right on the phone so you can test new mobile OS's on the go without plugging anything in. KDE plasma feels like a real mobile OS and worked surprisingly well. Updates for the entire OS are done from the alpine package manager. Building software on the phone felt liberating. My phone has a compiler and developer tools. I really hope the Pine Phone is my daily driver within a year or so.


I think the Nexus 5 has had the most attention historically, and may work best today. They have officially stated that the PinePhone is the future though - between the fact that it is easily available (to developers now, general availability expected in a couple months), and there is enough information that we think it is possible to make everything work it will soon be the best.

Nothing is currently usable for everything. PinePhone will soon be usable for phone calls, something nothing else can claim.



hm so one of the things least supported seems to be calls.. What can be done to make this happen? Should it happen?


Pass legislation that requires manufacturers of modem chipsets to open-source their firmware and driver blobs. That's the biggest obstacle.


We don't need open source blobs, we just need documentation on how to use them.

Open source blobs would be good for other reasons, but it isn't required to use them.


If all you have is a binary blob, doesn't that prevent you from ever recompiling it for use on a new OS or kernel?


That is normally irrelevant where it is true.

Some binary blobs are just firmware, and so you just need to know how to load it into the radio - this is generally trivial to port to something else. However you still need to know how to access the blob and that is often restricted.

Even where it needs to run in the local OS, it is physically soldered onto a board - nobody is going to put it in a different phone so the CPU is already know. As such it isn't hard to write an emulator for the parts of the host OS it needs. *BSD can run linux apps as if on a native linux kernel (they are generally a few years behind the latest linux kernel but that is typically good enough), and Wine allows running windows apps: the same thing can apply to blobs that need to run in your OS. However you need to know how to access the functions in the blob which is often not known.

Of course it is possible for a blob read/write to random places in memory in a way that works only for a specific kernel. If a blob does this it is very difficult to port to anything else (you have to figure out where it might write and ensure you never use that memory). This is rare and evil enough that if a blob is caught doing this kernel developers will make changes such that the blob won't run (blobs that do this tend to have bugs such that the entire OS is unstable and users blame the OS not the blob)


Doubt it. Frequency related laws mean it is unlikely any company would want to give the ability to modify anything that emit radio and microwaves.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: