It has been a problem that errors in the use of vectors and scalars, such as "x = 10 + vec(1,2,3)", did not generate an informative error message. Rather, such statements set x to "NaN", meaning "not a number", which caused errors when using x later in the program.
Now such errors give a message such as "Cannot add a scalar and a vector".
Also, if v was vector you couldn't say "v += vec(10,1,0)", but now you can. You can also use -=, *=, and /=.
Version 2.6 has been updated to provide these improvements.
There is one such issue that has not been addressed, which is that if v is a vector, v**2 is not caught as an error but gives the value "NaN".
Bruce