In trying to deeply understand continuity, I have found (to my limited knowledge) a disparity in the definitions of continuity at points and intervals.
By definition, a point is continuous if it’s left hand limit at x=a, right hand limit at x=a and f(a) are equal.
By definition, an endpoint is continuous if f(a) is equal to the limit which exists for that particular endpoint.
By definition, a function is continuous in an interval if the function is continuous at each point in the interval which in turn, is determined by the above definitions of continuity at points/endpoints.
Now, consider this diagram of a jump discontinuity:
We can state using the above definitions that f(x) is continuous in (- ∞, a] ∪ (a, ∞).
Combining the above ranges we can state that that f(x) is continuous in (- ∞, ∞). But now, again using the same definitions f(x) is discontinuous in (- ∞, ∞).
What have I missed?
The problem is that, for a function defined in $(-\infty, a]$, the limit at $x=a$ does not exist. We might thus make an extension of the definition of continuity to allow for continuity at the boundary. To do that, we demand instead that the left-handed limit is equal to $f(a)$, But note that this in an exception that is only made for points at the boundary. For every other point, we demand that both left and right-handed limits equal $f(a)$. Thus, when you consider the union $(-\infty, a]\cup(a,\infty)$ you cannot treat $a$ as an endpoint anymore. Since "continuity" at endpoints only demands side limits, you will end up with a function that has both a limit to the left of $a$ and a limit to the right of $a$, but they need not be equal.