uxor 
#matlab vs #gnuoctave : try computing
f=@(x,y) log(2+x)./atan(x-y)
c=@(x) -x
d=@(x) sin(x)
integral2(f,0,2,c,d)
then you'll get :
7.453231792656116 % with matlab
7.453230816525963 % with octave
a more accurate calculation gives
integral2(f,0,2,c,d,'AbsTol',1e-12)
7.453230816525963 % with matlab
7.453230816525963 % with octave
so matlab integral2 has not 1e-10 absolute error as said in the manual 🙄
before #gnuoctave version 9.* the command linsolve([-2,4,3;-2,-1,1],[0;0]) would lead to the obvious [0;0;0] solution like #matlab would do, but now gnuoctave give a non-null solution like null([-2,4,3;-2,-1,1]) would do. One more example to add to the list of differences between both softwares :
https://rouxph.blogspot.com/2024/10/divergences-entre-matlab-et-octave.html
