I am currently studying Adjunction spaces using Brown's Topology and Groupoids. I am having trouble understanding exercise 4.5.3:
Let $B$ be a closed subspace of $Q$. For each $λ=1,\dots,n$, let $f_λ:X_λ→Q$ be a map, and let $A_λ$ be a closed subspace of $X_λ$ such that
- $f_λ[A_λ]⊆B$,
- $f_λ|X_λ\setminus A_λ$ is injective,
- the sets $f_λ[X_λ\setminus A_λ]$ are disjoint and cover $Q\setminus B$,
- $f_λ|X_λ$, $f_λ[X_λ]$ is an identification map.
Prove that a function $g:Q→Y$ is continuous if and only if $g|B$ ,$gf_λ$ is continuous, $λ=1,\dots,n$. Prove also that there is a homeomorphism $Q→B_{f1}⊔X1\dots_{fn}⊔Xn$ which is the identity on $B$.
I tried to prove that $Q$ has the final topology with respect to the inclusion $i_B:B\rightarrow Q$ and the maps $f_\lambda$. For that I take a subset $C$ in $Q$ whose inverse image for all these maps are closed, and I try to show that $C$ is closed. However, I am not able to prove it without adding the assumption that $f_\lambda[X_\lambda]$ is closed in $Q$ for each $\lambda$, and I have the feeling that it is necessary.
For example, take $Q=[0,2]$ and the closed subspace $B=[0,1]$ in $Q$. Take $X=(0,1)\cup (1,2]$ and the closed subspace $A=(0,1)$ in $X$. Let $f:X\rightarrow Q$ be the identity map. Then these spaces and maps satisfy the conditions above (do they, actually?). But the subset $C=(1,2]$ of $Q$ is not closed in $Q$ although $i_B^{-1}[C]$ is closed in $B$ and $f^{-1}[C]$ is closed in $X$. I cannot find where is the mistake in this example.
How can I solve this exercise without the additional assumption? Is my example wrong?