I had to calculate:
$$\int_0^\infty\textrm{d}x\,\frac{1-\exp(-t)}{t}\cos(t)$$
My approach was to define a function with parameter:
$$\mathcal{F}(a)=\int_0^\infty\textrm{d}x\,\frac{1-\exp(-a t)}{t}\cos(t):=\int_0^\infty\textrm{d}x\,f(x,a)$$
Then we can consider:
$$\partial_af(x,a)=\exp(-a t) \cos(t)$$
Which is clearly dominated for $a\in[a_0,\infty[$ (for $a_0>0$) by a function: $\exp(-a_0t)$. Then there is a theorem that states that if there exists some $a_1$ for which:
$$\mathcal{F}(a_1)\; \textrm{is well defined} \quad \textrm{and}\quad \partial_af(x,a)\;\textrm{is dominated}$$
Then $\mathcal{F}(a)$ can be redefined to:
$$\mathcal{F}(a)=\mathcal{F}(a_1)+\int_{a_1}^{a}\textrm{d}y\int_0^\infty\textrm{d}x\;\partial_yf(x,y)$$
Which is then well defined on $a\in[a_0,\infty[$.
Then by exact calculation:
$$\int_0^\infty\textrm{d}x\;\exp(-a t) \cos(t)=\frac{a}{1+a^2}$$
And now integrating I finally got expression:
$$\mathcal{F}(a)=\frac{1}{2}\log(1+a^2)+C$$
And here emerges the problem to find the value of $C$. Then the final integral would be for $\mathcal{F}(1)$.
With Mathematica I got that $\mathcal{F}(1)=\frac{1}{2}\log(2)$ therefore $C=0$ and if someone could prove that:
$$\lim_{a\to0}\int_0^\infty\textrm{d}x\;f(x,a)=\int_0^\infty\textrm{d}x\;\lim_{a\to0}f(x,a)$$
I would be very grateful, because then $\lim_{a\to0}\;f(x,a)=0$ and $C=0$.
For any $a_1>0$, we get from your formulas
$F(a) = F(a_1) + \frac{1}{2}\ln(1+a^2) - \frac{1}{2}\ln(1+a_1^2)$.
Now letting $a_1 \to 0$, using continuity of $F(a)$ and obvious fact that $F(0)=0$, we get the following.
$F(a) = \frac{1}{2}\ln(1+a^2)$
Hence, $F(1)=\frac{1}{2}\ln(2)$.