On ncatlab, in explaining the basics of the Salamander lemma, they state the following.
For $A_{\square}$, they write the denominator as $\operatorname{im}(\partial^{hor}_{in}) \oplus \operatorname{im}(\partial^{vert}_{in})$. Why are they using the direct sum here? Those two modules aren't disjoint, are they?
