# \[ANN\] LightGraphs v1.3.3 - another bugfix

**URL:** <https://discourse.julialang.org/t/ann-lightgraphs-v1-3-3-another-bugfix/39191>\
**Category:** Package Announcements\
**Tags:** lightgraphs\
**Created:** [May 9, 2020, 6:59pm UTC](https://discourse.julialang.org/t/ann-lightgraphs-v1-3-3-another-bugfix/39191 "2020-05-09T18:59:25Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![anon94023334](https://avatars.discourse-cdn.com/v4/letter/a/e274bd/32.png) [@anon94023334](https://discourse.julialang.org/u/anon94023334)\
**Post date:** [May 9, 2020, 6:59pm UTC](https://discourse.julialang.org/t/ann-lightgraphs-v1-3-3-another-bugfix/39191/1 "2020-05-09T18:59:26Z")

</div>

Hi all,

LightGraphs.jl has been updated to v1.3.3 to fix a bug in `core_number`. No other changes have been made.

I expect to have a cool LG announcement in the next few days. Please stay tuned 😃.

Edited to add more detail:

This is a bug in the _correctness_ of `core_number`. The old version produced incorrect results when run on directed graphs with bidirectional edges. The reason is that the old code used `all_neighbors` to determine the set of neighbors to process; the issue is that `all_neighbors` ensures no duplicates: so if you had an edge from X → Y and another from Y → X, the neighbors of X would only include Y one time. This is not what the algorithm specifies.

So we fixed it by explicitly using `outneighbors` and `inneighbors` so that bidirectional edges are included. Fortuitously, this also resulted in a speedup as `all_neighbors` needed to deduplicate; and we explicitly do not want that behavior here.

So - if you’ve been using `core_number` results with directed graphs anywhere, (including with `k_core` and the other degeneracy functions), you will probably want to re-run your results with this fix applied.