os: Enable GPS on adafruithat only

[sonnyp]

The GPS hat is not tested and probably incompatible with PlanktoScope hat.

[sonnyp]

This will be a blocker for #583

We will have 3 options

Remove GPS support

Do people use the GPS hat? It seems like a lot of related things were removed over time.

Remove adafruithat support

And make the release before #583 the latest release to support it

I understand a lot of PlanktoScope with adafruithat are out there and being used so it seems premature. Maybe we can find a way to

• make maintenance and testing easier
• call for help

I have the project to have automated test suites that run on real hardware, we could add an adafruithat to the test bench.

Change how hardware configuration works

See #583 (comment) for context

If we extend device-portal with hardware configuration we can do hardware/software setup at runtime rather than build time but this will be significant work

• scripts will need to be atomic (as much as possible)
• scripts will need to be idempotent
• scripts will need to have install/remove steps

On the other hand, it would be good to have this for developer experience anyway.

[ethanjli]

Here are some relevant comments from my notes for past meetings:

• 2023-10-19 software meeting: according to @chevreuill3000, "the GPS recorded by the machine doesn’t correspond to the location used for sampling - it’s basically useless for GPS location."
• 2023-08-31 software meeting: according to @fabienlombard, "we don’t need to worry too much about GPS feature - currently it’s 100% useless. It could be useful if the PlanktoScope was able to actually acquire unfiltered water so that we could use it on an inline system (like IFCB). What is more important is to have an accurate system time."

To add some nuance/context to my summaries of these comments:

• The GPS location of the PlanktoScope sometimes (but does not always) correspond to the location where samples were obtained. If a sample was collected in one place and then imaged later with a PlanktoScope at a different place, the PlanktoScope's GPS coordinates are completely useless. On the other hand, if a sample was collected and then imaged immediately on-site at the same location, the GPS coordinates are useful. I don't know how often the former use-case occurs vs. the latter use-case - maybe @kitazy52 will have some relevant knowledge based on usage patterns in Japan.
• One of the major projects for future PlanktoScope functionality is to add a capability for using the PlanktoScope on an inline system (so that the PlanktoScope is continuously imaging samples exactly when/where it is collecting the water). I believe @babo989 and @chevreuill3000 are involved in conversations regarding such plans. I don't know how GPS-related information will be collected and recorded in such a scenario - but I assume that if a GPS module is included in future PlanktoScope systems, the GPS module will be not be integrated via Adafruit's GPS HAT.

Removing GPS support would also help make the Adafruit HAT and PlanktoScope HAT versions of the Node-RED dashboard more similar to each other in preparation for a future unification (though if the v3 dashboard is completed first, then maybe we don't need to worry about unifying our current/v2 dashboards).

Based on the above information, I think it's reasonable to deprecate GPS support in v2025.0.0. After I finish #590, I think the main user impact of removing GPS support will be to require manual user input of GPS coordinates, which I think is an acceptable impact. However, it would be nice to get an updated second opinion from users who rely on the Adafruit HAT (e.g. @kitazy52 ) so that we can have a chance to be made aware of any other potential impacts we haven't considered thus far.

If we want to fully remove GPS support in v2025.0.0, I think we should have a discussion to weigh potential impacts of that removal together with the general rule I've been following for PlanktoScope OS release planning that "In general, user-facing parts of the software should only be removed if they were deprecated in a previous YYYY.minor release.". Maybe we'll end up deciding to relax that rule in this specific case or in general...I'm open to possibilities.

If we want to include #583 in v2025.0.0 without a full removal of GPS support in v2025.0.0, what I would propose in v2025.0.0 is to install the necessary programs for GPS support in the OS image, but leave the relevant systemd services disabled, and provide a (deprecated-at-birth) enablement script in the OS image; then the user could make the necessary changes to enable GPS support (even without internet access) by running the enablement script in the Cockpit Terminal, similar to the UX for enabling developer mode. In other words, for v2025.0.0, GPS support would completely be disabled by default (but still accessible by opt-in action), as part of v2025.0.0's complete deprecation for GPS support; this will also help us get some information on whether any users are relying on the GPS HAT in practice, based on the amount (or absence) of users asking for help. Then for a subsequent release of PlanktoScope OS, GPS support could completely be nonexistent on the OS image: we wouldn't install any of the necessary programs when we build the OS image, and we wouldn't provide the enablement script either.

[sonnyp]

If we want to include #583 in v2025.0

Not from my side and it wouldn't be ready on time anyway.