comment #4:

Ahh, I originally misunderstood your phrase in your comment #1 that "in the meantime, the B = sum(A, "all") behavior can be emulated in Octave and pre 2018 Matlab using B = sum(A(:))". I thought you meant that I couldn't replace first with second in Matlab after the 2018+ version. But you really mean that before Matlab 2018 I can't use these syntax 'sum(A, "all")'.

No, never mind, please...

comment #2:

> just realized that in comment #! i meant to point to the short list in bug #58116 comment #3.

Thank you for your links, Nicholas! Unfortunately, I couldn't apply this solution with 'B = sum(A(:))' because I and my colleagues use Matlab 2020, although it could be fixed by using appropriate number of sums, e.g.: B = sum(sum(A)).

P.S. At this point I concluded that we cannot use Octave yet, not even because of the lack of support for summation on all axes (this can be easily fixed), but because it does not have important functions for our program, namely the 'yyaxis' (bug #63108) (in short, function that plot graphics with 2 independent y axes)

P.P.S. I am a C++ programmer and could probably add support for 'all' parameter to some of listed in bug #58116 functions, but unfortunately I don't have time to do that yet.

there have been a few different reports about the changes made to those functions in 2018. in addition to 'all' (bug #58116), there is the ability to take vector dimensions (bug #58089), and the nanflag option (bug #50571).

regarding other functions, i made a short list bug #58114 comment #3, and I think the change overlaps with the functions that accept nanflag, which has a more extensive list in bug #50571.

closing as a duplicate, further discussion can be added to the existing reports.  in the meantime, the B = sum(A, "all") behavior can be emulated in Octave and pre 2018 Matlab using B = sum(A(:))