Consider $\gamma\colon [0,2\pi]\longrightarrow \mathbb{C}, \ \gamma(t)=3e^{it}$
I want to compute $\int_\gamma \frac{1}{(z+2)^n} \, \mathrm{d}z$.
I am stuck with $\int_0^{2\pi}\frac{3ie^{it}}{(3e^{it}+2)^n} \, \mathrm{d}t$ where it is tempting to substitute, but this can not be done that easily, as the substitute would also require some sort of path. If I was just naively substituting the result would be wrong. How can this issue be resolved, and this is important, without using Cauchy formular nor any advanced theorem?
You still can use the fundamental theorem of calculus. $\int_\Gamma g(z)dz$ for some path $\Gamma$ is equal to the change of $f(z)$ along the path, where $f(z)$ is a primitive of $g(z)$, i.e. $f'(z)=g(z)$. In your case, assuming $n$ is a positive integer:
$$f(z) = \begin{cases} \frac{-1}{(n-1)(z+2)^{n-1}}, &\text{if $n>1$}\\ \ln(z+2), &\text{if $n=1$} \end{cases}\;\Longrightarrow\; f'(z)=\frac{1}{(z+2)^n}\,. $$
Because $f(z)$ is single valued for $n>1$, and your path is closed, the integral is zero if $n>1$ ($f(z)$ does not change after the path is done). For $n=1$ the change of $\ln(z+2)$ along you path is $i$ times the change of the phase of $z+2$ along the loop, which is $2\pi$, so the integral has a value of $2\pi i$. The big difference in the $n=1$ case is that $f(z)$ is multivalued, so when you return to the original point in the loop $f(z)$ has increased by $2\pi i$.