Help with ERA errors

ChatGPT suggests that dc.getWidth() or dc.getHeight() can return 0 in onLayout(). Is this true?

That wouldn't make sense in 2 ways: Why would Garmin have an invisible size? And more importanty that wouldn't cause a crash.

Are you sure that the code, line numbers we see in github are whats's causing it? Can it be that you had some changes in the code since or maybe you hade some comments in the code when you made the build that moved the lines?

My suspect would be the Match functions. While the documentation doesn't say anything about the range of the input values, can that cause this?

flocsy🤠 said:

Are you sure that the code, line numbers we see in github are whats's causing it? Can it be that you had some changes in the code since or maybe you hade some comments in the code when you made the build that moved the lines?

The link to the github points to the exact revision I used to build version 2.0. Maybe there is some inaccuracy in the line numbers when the exception is raised, I don't know.

flocsy🤠 said:

My suspect would be the Match functions. While the documentation doesn't say anything about the range of the input values, can that cause this?

This doesn't explain why the code works in 99.99% cases, but sometimes doesn't.

True. But we also don't really know any statistics.

flocsy🤠 said:

True. But we also don't really know any statistics.

Developer Dashboard reports 1K+ Downloads, and 21% users installed version 2.0, that gives at least 200 onLayout() calls And there are 7 crashes in ERA.

With what you see in ERA, it only a specific or language?  With 200 crashes, that could be a low numbers of users that keep seeing it, (20 users that saw it 10 times each for example)  What you see in ERA are not unique users. and looking at specific device/FW versions/languages can help sort that out.

jim_m_58 you just have to look in the original post:
Languages: eng, ita
Languages: eng, por

but the code doesn't look like there's anything language related.

I once had an error, that was something with Session.createField, and I had lots of errors, but all were in Thai devices. It turned out that the Thai translation of the string that was used in the :units was too long:  documentation change request: remove or elaborate on requirement for current device language for :units in Session.createField 

 feature-request: add maxLength, maxSize attributes to string resources and fail at compilation if a string is too long 

This may or may not help, but you say 'simple arithmetic operations' but you could [for the purposes of this] simplify them more by splitting the VARs with 3 operations into two vars:

i.e. da_height = 2 * dig_height + dy;

becomes

da_heightB = dig_height + dy;

da_height = 2 * da_heightB;

and do the same with the lines that are being reported i.e. xd[0] = (dc_width - da_width) / 2;

This might help you determine what / where specifically the issue is.

The other factor that you are not thinking about is this is Garmin / Monkey C / CIQ so there might not be any logic to the errors, and the errors might not actually be real to the user (and the watch face loads and is displayed properly to the user). And lastly it is only a maximum of 3.5% of users watches that are reporting the errors, which in the real world would be very bad but maybe not so much in this world.