[chirp_devel] Comments on a UI rewrite

[Dan Smith]

> It will become increasingly difficult to support software on top of OSes, which themselves are abandoned, especially as underlying software packages and dependencies have moved on. To continue supporting this would likely mean all sorts of unsavory acrobatics and ugly hacks in Chirp.
> I think it's reasonable to set this line to be currently supported OSes, e.g. OSX High Sierra+, Windows 7+, etc.
> In fact, Windows 7 is reaching end of life in a matter of weeks! (It seems reasonable to continue supporting it for near future, as it has a big enough base.)
> Nothing stops folks from continuing use of older versions of Chirp on existing radios, but for new feature and new driver support, they will need to use a modern (in a very loose sense of the word) OS.

Obviously we can't support everything forever, and nobody would suggest that. WinXP is similar enough to modern times that supporting it has, thus far, not been difficult (at all), and runs on a lot of old hardware. It's not that I don't think we should ever move the minimum version forward, it's just that I see no reason to do it until or unless there is a good reason to do so.

> On a side note, Derek mentioned something interesting: usage stats. Does chirp have install/usage telemetry? I.e. can you actually see where and how the software is getting used, in the real world? Else - not having seen any real data myself, and pure speculation - I would imagine you have a very small, but vocal number of XP, etc users who want you to continue supporting them through the apocalypse.

Yes, I have some stats. However, they don't count users that aren't connected at the time of use. Given how it's difficult to even get a modern browser on XP and navigate the https world, I suspect most of the people using it are doing so by downloading chirp elsewhere and using it on XP offline. And, of course, not everyone allows the reporting.

A quick query out of the stats show that in 2019, 3% of all windows users were on XP. That was 4% in 2018 and 6% in 2017. All versions of MacOS constituted about twice the raw number of WinXP reports in 2019. Clearly a downward trend indicating that the number of people affected is shrinking, as MacOS users are the smallest fraction. MacOS and Linux users (combined) constitute about 10% of the total, so choosing to keep 2% of users on XP instead of the 10% on Mac/Linux seems like an obviously bad idea.

Either way, dropping WinXP was lowest on my list of concerns (the list in the top of this thread wasn't really sorted), and the stats certainly support that not being as import

> Regarding contribution on the driver migrations, I haven't been actively contributing in a long time, but I think a move to modernize/revamp Chirp might spur me on to revisit. (I still have several Yaesu radios that I can develop and test against). Is there a documented guide/wiki/etc on how to do this migration, or just pointing to examples of stuff already been migrated? (At least a conceptual outline may be a useful starting point for figuring out how to attack the migration of a given driver.)

No, but the tests are a good place to start. Fix the syntax issues for py3 so the tests will import the driver, and then iterate until it passes. And, of course, looking at the diffs of conversions of other drivers. We could start a wiki page with issues and resolutions, but I haven't done that yet.

> This might be a good time to rekindle an old topic I have brought up before, as I saw another recent thread mentioning Github as well.

I do not want to change the ui toolkit, the language, and the entire process all at once, no. The thing I value the most is the long-time contributors that have done most of the work here, many of which are not professional developers. I understand the desire to move towards a github workflow, but it's just not that important to me. I (and other devs here) have tooling and processes setup for how things are now. The python3 thing is an acute problem needing solving -- the process one is not at the moment.

--Dan