Representative devices for testing

What would be a good representative sample of devices for testing, so that I could reasonably expect a watch face to work on all or nearly all devices without having to test all 90+ of them?

I have come up with this list. It covers most of the major.minor versions, screen types and shapes, and color depths:

displayName	deviceId	displayType	deviceFamily	bitsPerPixel	audioContentProvider _memoryLimit_	background _memoryLimit_	datafield _memoryLimit_	glance _memoryLimit_	watchApp _memoryLimit_	watchFace _memoryLimit_	widget _memoryLimit_	partNumbers_0_ _connectIQVersion_
Forerunner® 235	fr235	mip	semiround-215x180	4			16384		65536	65536	65536	1.4.4
vívoactive® HR	vivoactive_hr	mip	rectangle-148x205	8		32768	32768		131072	65536	65536	2.4.1
fēnix® Chronos	fenixchronos	mip	round-218x218	8		32768	32768		131072	98304	65536	3.1.6
D2 Air	d2air	amoled	round-390x390	16	524288	65536	32768		1048576	524288	524288	3.2.1
Instinct® 2S / Solar / Dual Power	instinct2s	lcd	semioctagon-163x156	1		32768	32768	32768	98304	65536	65536	3.3.4
fēnix® 7X / tactix® 7 / quatix® 7X Solar / Enduro 2	fenix7x	mip	round-280x280	8	524288	65536	262144	65536	786432	131072		4.1.2
Venu® 2 System 6 Preview	venu2system6preview	amoled	round-416x416	16	524288	65536	262144	65536	786432	131072		4.2.0

I have a Forerunner 235 and Fenix 7x so I can do actual hardware testing which is why these two are on the list. I know the simulator isn't always great so does it do a decent job with these? Any more I should consider, or is one of these not a good choice for some reason?

0 flocsy🤠 over 2 years ago

I think it depends on your project. Looking at mine projects the answer is that if you analyze your code (probably monkey.jungle is enough, thou if you use "has" to detect features then you'll need to factor that in)

What I do is this:

1. Gather the CIQ version that "matter". What I need on this is that for example the app uses different code on devices that don't have CIQ 2.4.0, so this version is interesting. But other part of the code behaves different if the device has CIQ 3.2.0. etc..

2. Gather form factors (if applicable): square, round, maybe even resolution

3. Gather used, but optional features. I.e if device has barometer.... Color screen vs B&W

4. In some cases the available memory size (i.e for datafields)

5. After gathering all these requirements you need to categorize every device. This is not easy, but usually by the time you get to ask the question you asked you have a pretty good idea, as I wrote usually the monkey.jungle is evolving to this direction anyway

6. Now you can pick one device from each group