ndarray v0.16.1
|-- argmin-math v0.5.1
| |-- argmin v0.11.0
| | `-- rssn v0.1.13
| `-- rssn v0.1.13
|-- ndarray-npy v0.9.1
| `-- rssn v0.1.13
|-- rssn v0.1.13
`-- sprs-rssn v0.11.4
`-- rssn v0.1.13
This is the way we solve this issue for now.
argmin explicitly uses ndarray v0.16.1 but sprs uses 0.15.0 to 0.18.0.
That caused cargo to choose ndarray v0.17.1 for sprs but our main library uses 0.17.1.
We do not want to have two version of ndarray here (just like what we do with rand) because it's too big so we just fork sprs and limited it to use 0.16.1.
And that is the issue: We have to "merge“ changes from sprs frequently or find another solution.
This is the way we solve this issue for now.
argmin explicitly uses ndarray v0.16.1 but sprs uses 0.15.0 to 0.18.0.
That caused cargo to choose ndarray v0.17.1 for sprs but our main library uses 0.17.1.
We do not want to have two version of ndarray here (just like what we do with rand) because it's too big so we just fork sprs and limited it to use 0.16.1.
And that is the issue: We have to "merge“ changes from sprs frequently or find another solution.