Slightly modified test code which is easier to see that both plots end up in the same axes is

H1 = subplot(2,1,1);
H2 = subplot(2,1,2);
stairs(H1, 1:5,1:5, 'b')
stairs(H2, 1:5,5:-1:1, 'r')

Script is attached as tst_stairs.m

(file #53166)

Re-creating a bug report for the issue reported in patch #10200 so that it can be marked as a blocker for Octave 7.2.

The original report:
stairs ignores first arg: axes handle

H1 = subplot(2,1,1);
H2 = subplot(2,1,2);
stairs(H1, 1:5,1:5, 'b')
stairs(H2, 1:5,1:5, 'r')

both draw commands goes into last axes handle
This does not happen with plot(H, ...)


Addendum: This is a regression from bug #61945 for which `newplot` was changed to no longer set the current axes. (I pushed those changes probably a little to hastily.)
This does not only happen for `stairs` but for all plot functions that expect `newplot` to change the current axes. That is most probably true for functions that combine `newplot` with `__next_line_color__.m` and `__next_line_style__.m` (because the latter always act on the current axes).

Functions that combine those functions and might therefore be affected are:
- area.m
- plot3.m
- stairs.m
- bar.m
- stem.m

To be save, we should probably check all functions that use `newplot` (which are many)...

Should we revert the changes for bug #61945 and work out a proper fix for this before reapplying them on default?