TODO in unweighted shortest path implementations (BFS)?

szhorvat · 3 March 2020 09:25

@Gabor Do you recall why this is marked as TODO? I don’t see what’s wrong. There are several instances of the same and I am working with this code at the moment.

github.com

igraph/igraph/blob/master/src/structural_properties.c#L252


*res = 0;
if (directed) {
    dirmode = IGRAPH_OUT;
} else {
    dirmode = IGRAPH_ALL;
}
already_added = igraph_Calloc(no_of_nodes, long int);
if (already_added == 0) {
    IGRAPH_ERROR("average path length failed", IGRAPH_ENOMEM);
}
IGRAPH_FINALLY(free, already_added); /* TODO: hack */
IGRAPH_DQUEUE_INIT_FINALLY(&q, 100);


igraph_adjlist_init(graph, &allneis, dirmode);
IGRAPH_FINALLY(igraph_adjlist_destroy, &allneis);


for (i = 0; i < no_of_nodes; i++) {
    nodes_reached = 0;
    IGRAPH_CHECK(igraph_dqueue_push(&q, i));
    IGRAPH_CHECK(igraph_dqueue_push(&q, 0));
    already_added[i] = i + 1;

Gabor · 3 March 2020 09:42

Because it assumes that igraph_Free() is the same as free().

igraph_Free() is a macro, so you cannot take its address, but we should also define igraph_Free() as a function, in addition to the macro, and then we could use it here as well.

szhorvat · 3 March 2020 10:00

We already have the function igraph_free(). I’ll change it to that.

Topic		Replies	Views
C/igraph 0.10.10 Announcements C	0	87	13 February 2024
C/igraph 0.8.3 Announcements	0	638	3 October 2020
C/igraph 0.9.7 Announcements	0	285	16 March 2022
Implementing the LDBC Graphalytics benchmark Usage C	14	1127	6 February 2021
C/igraph 0.10.4 Announcements C	0	219	27 January 2023

TODO in unweighted shortest path implementations (BFS)?

Related topics