Alles anzeigenI know this sounds logical but I would love to know how much it actually is. And I admit I have no real way to find out, I'm not all-knowing after all. I know that the string resources are all parsed and saved on the heap by android.content.res.AssetManager, which uses cpp code (made for JNI, so it does create Java objects in the end). From what I can see these don't retain the original XML structures at all, and instead save in "ResTables" that use maps with resource ids to resources, and those resource IDs are implicitly created integers in our resource namespace anyway. So from what I can tell from reading Android code they don't look "horrifying", and languages or resource files that are not used aren't parsed at all.
Last time we discussed ram you included shared memory allocations in the "ram usage" of the application, shared memory that is shared system wide, making 64 bit memory usage insanely blown up. I think there needs to be a bit of a higher standard in seeing what one "probably assumes" to be the case as the true facts and that your own code is always better than what these weird Californian Google dudes can produce.
For translations, here's a few things I found
Alles anzeigenCode
< constexpr static std::string_view delabels[]={ "Ins schnell", "Kohlenhydra", "Dextrose", "Langes Insu", "radeln", "Walk", "Blut"};> constexpr static std::string_view delabels[]={ "Schnellinsulin", "Kohlenhydrate", "Dextrose", "Verzögerungsinsulin", "radeln", "Walk", "Blut"};> .monthlabel={ "Jan", "Feb", "Mär", "Apr", "Mai", "Jun", "Jul", "Aug", "Sep", "Okt", "Nov", "Dez"},< {"Fehler bei Bibliotheksinitialisierung", "Neuinstallation durch Entfernen der Bibliothek erzwingen"},
You seem to think that because Android is made a certain way, it has to be efficient. That is really stupid. Do you want to say that the Libre3 app is smaller and faster than Juggluco? The reason they make it the way they do has more to do with other concerns than efficiency:
First I naturally came at "Benachrichtigen", but that is way too long, if Benachricht is really impossible German I have to use English. I also had "Mär" instead of "Mä", but that uses 5 bytes instead of the 4 I used in the multidimensional array. In last version I have just changed the array to 5 bytes.
I also didn't abbreviate these example labels without a reason; labels are maximally 12 bytes including a terminating 0. They otherwise also don't fit in the Kerfstok watch app. This can't be changed easily; these structures already sit in installed programs. Also the space after "Dunkelmodus" is there for a reason, [x] should be placed after it.
For the help and strings.xml, I don't have those restrictions.