I often saw a 'proof' that $\int_1^\infty\frac{\sin(x)}{x}dx$ converges:
By integration by parts we get $$\int_1^\infty\frac{\sin(x)}{x}dx = \cos(1)-\int_1^\infty{\frac{\cos(x)}{x^2}}dx$$ and thus because $\frac{\cos(x)}{x^2} \leq \frac{1}{x^2}$ and $\int_1^\infty{\frac{1}{x^2}}dx$ converges $$\int_1^\infty{\frac{\cos(x)}{x^2}}dx$$ also converges by comparison test.
But isn't it just wrong to reason like that? Just because $$\lim_{C\to\infty}\int_{1}^C{\frac{1}{x^2}dx}$$ exists and $$\int_{1}^C{\frac{\cos(x)}{x^2}dx} \leq \int_{1}^C{\frac{1}{x^2}dx}$$ for all $C\geq 1$, it dosn't follow that $$\lim_{C\to\infty}\int_{1}^C{\frac{\cos(x)}{x^2}dx}$$ exists. Am I missing something obvious? This 'proof' also appears in the book "Analysis 1" by Königsberger.
Maybe the clearest intuition comes from sequences. Let $(c_n)$ be any increasing sequence which goes to infinity, and consider $$I_n=\int_1^{c_n} \frac{\cos x}{x^2}\,dx.$$ We have $$|I_n-I_m|\le \int_{c_m}^{c_n} \frac{1}{x^2}\,dx,$$ so the sequence $(I_n)$ is Cauchy, and therefore converges.