Doc and news for dropping scalar dimensions

NEWS.md

@@ -37,10 +37,15 @@ Library improvements
 
   * New method for generic QR with column pivoting ([#13480]).
 
-  * A new `SparseVector` type allows for one-dimensional sparse arrays. Slicing
-    and reshaping sparse matrices now return vectors when appropriate. The
-    `sparsevec` function returns a one-dimensional sparse vector instead of a
-    one-column sparse matrix.
+  * Arrays:
+
+    * A new `SparseVector` type allows for one-dimensional sparse arrays. Slicing
+      and reshaping sparse matrices now return vectors when appropriate. The
+      `sparsevec` function returns a one-dimensional sparse vector instead of a
+      one-column sparse matrix.
+
+    * All dimensions indexed by scalars are now dropped, whereas previously only
+      trailing scalar dimensions would be omitted from the result.
 
 Deprecated or removed
 ---------------------

doc/manual/arrays.rst

@@ -233,8 +233,8 @@ where each ``I_k`` may be:
 The result ``X`` generally has dimensions
 ``(length(I_1), length(I_2), ..., length(I_n))``, with location
 ``(i_1, i_2, ..., i_n)`` of ``X`` containing the value
-``A[I_1[i_1], I_2[i_2], ..., I_n[i_n]]``. Trailing dimensions
-indexed with scalars are dropped. For example, the dimensions of ``A[I, 1]`` will be
+``A[I_1[i_1], I_2[i_2], ..., I_n[i_n]]``. All dimensions indexed with scalars are
+dropped. For example, the result of ``A[2, I, 3]`` will be a vector with size
 ``(length(I),)``. Boolean vectors are first transformed with ``find``; the size of
 a dimension indexed by a boolean vector will be the number of true values in the vector.
 As a special part of this syntax, the ``end`` keyword may be used to represent the last