This has been asked before, but I am looking for a proof that does not use integration by parts. Instead, I am looking to use what my textbook called the "Identity Criterion", which states that if the derivative of two functions are equal, then they differ at most by a constant.
Here is what I have thus far: By the Second Fundamental Theorem, $$f'(x) = \int_0^x f''(t)dt.$$ Adding something inconspicuous to the right side, $$f'(x) = xf''(x) + \int_0^x f''(t)dt - xf''(x) = \frac{d}{dx} \left[ x\int_0^x f''(t)dt - \int_0^x tf''(t)dt \right] = \dfrac{d}{dx} \left[ \int_0^x (x-t)f''(t)dt \right].$$ By the Identity Criterion, $f(x)$ and $\int_0^x (x-t)f''(t)dt$ must only differ by a constant. But by the First Fundamental Theorem, $$\int_0^x (x-t)f''(t)dt = $$
This is where I start to lose the plot. Any guidance would be useful.
Why not just integration by parts?
\begin{align*} f(x)-f(0)&=\int_{0}^{x}f'(t)dt\\ &=\int_{0}^{x}-\dfrac{d}{dt}(x-t)f'(t)dt\\ &=-(x-t)f'(t)\bigg|_{t=0}^{t=x}+\int_{0}^{x}(x-t)f''(t)dt\\ &=xf'(0)+\int_{0}^{x}(x-t)f''(t)dt. \end{align*}
Edit:
Let \begin{align*} \varphi(x)&=f(x)-\left(f(0)+f'(0)x+\int_{0}^{x}(x-t)f''(t)dt\right)\\ \varphi'(x)&=f'(x)-f'(0)-\int_{0}^{x}f''(t)dt-xf''(x)+xf''(x)\\ \varphi'(x)&=f'(x)-f'(0)-(f'(x)-f'(0))\\ \varphi'(x)&=0, \end{align*} and so $\varphi(x)=c$, but $\varphi(0)=0$ and hence $c=0$.