Closed simonschoelly closed 3 years ago
Merging #1514 (e5595b2) into master (ce53372) will increase coverage by
6.73%
. The diff coverage is55.55%
.
@@ Coverage Diff @@
## master #1514 +/- ##
==========================================
+ Coverage 92.66% 99.39% +6.73%
==========================================
Files 195 107 -88
Lines 6325 5156 -1169
==========================================
- Hits 5861 5125 -736
+ Misses 464 31 -433
if we add this to a patch version, we can hope folks will upgrade their LG
Maybe we could just enforce a new enough version of LightGraphs in Project.toml.
This is the most correct way of doing this. I'll commit this bugfix and then tag a new release. Could you please make a PR to MetaGraphs Project.toml to specify the new release? Then we can tag a new MetaGraphs release.
Also, generally speaking, should we implement hash
on other edge types (like SimpleWeightedEdge
)?
Thanks!
Previously the
==
method was overriden forAbstractSimpleEdge
but nothash
. In order for dictionaries for edges to work correctly, we also have to implementhash
so thate1 == e2
implieshash(e1) == hash(e2)
.This has some serious consequences for
MetaGraphs.jl
, whereweights(g)
alawys returns values of1.0
whenever the eltype of aMetaGraph
is notInt
. See the example below:This PR should fix that, but I was wondering if we also should have some fix in
MetaGraphs
to ensure that no one produces incorrect results whenever they use a not up-to-date version ofLightGraphs
. Maybe we could just enforce a new enough version ofLightGraphs
inProject.toml
.