From aa3d19767a280520bbc305fab7404867537efd8f Mon Sep 17 00:00:00 2001
From: Guillaume Dalle <22795598+gdalle@users.noreply.github.com>
Date: Thu, 18 Jul 2024 09:13:44 +0200
Subject: [PATCH] A star weights warning (#392)

---
 src/shortestpaths/astar.jl | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/src/shortestpaths/astar.jl b/src/shortestpaths/astar.jl
index c1c799a47..d42bd442c 100644
--- a/src/shortestpaths/astar.jl
+++ b/src/shortestpaths/astar.jl
@@ -70,6 +70,9 @@ Return a vector of edges.
 - `distmx::AbstractMatrix`: an optional (possibly sparse) `n × n` matrix of edge weights. It is set to `weights(g)` by default (which itself falls back on [`Graphs.DefaultDistance`](@ref)).
 - `heuristic`: an optional function mapping each vertex to a lower estimate of the remaining distance from `v` to `t`. It is set to `v -> 0` by default (which corresponds to Dijkstra's algorithm). Note that the heuristic values should have the same type as the edge weights!
 - `edgetype_to_return::Type{E}`: the type `E<:AbstractEdge` of the edges in the return vector. It is set to `edgetype(g)` by default. Note that the two-argument constructor `E(u, v)` must be defined, even for weighted edges: if it isn't, consider using `E = Graphs.SimpleEdge`.
+
+!!! warning
+    Since a two-argument edge constructor `E(u, v)` is used to construct the path, metadata associated with the edge (like its weight) will be lost in the result. You might need to code a post-processing step yourself.
 """
 function a_star(
     g::AbstractGraph{U},  # the g