We use a relatively small set of natverse tools within CATMAID and constructing a full R environment to do so is slow (as most dependencies need to be compiled) and error-prone (PAT token rate limiting, build time dependencies etc.). Then the code is tricky to read and maintain compared to a pure python solution like navis. And it would be nice to avoid CATMAID's backend standing up a client which then uses HTTP to connect to itself.
As I understand it, we primarily use this for transformations and NBLAST, both of which exist already in navis. If navis needs any additional functionality (or we need any additional adapters for the thin wrappers around existing datasets like fafbseg) I'm sure @schlegelp would be happy for us to contribute.
We use a relatively small set of natverse tools within CATMAID and constructing a full R environment to do so is slow (as most dependencies need to be compiled) and error-prone (PAT token rate limiting, build time dependencies etc.). Then the code is tricky to read and maintain compared to a pure python solution like navis. And it would be nice to avoid CATMAID's backend standing up a client which then uses HTTP to connect to itself.
As I understand it, we primarily use this for transformations and NBLAST, both of which exist already in navis. If navis needs any additional functionality (or we need any additional adapters for the thin wrappers around existing datasets like fafbseg) I'm sure @schlegelp would be happy for us to contribute.
It looks like this is all of the R interface stuff https://github.com/catmaid/CATMAID/blob/dev/django/applications/catmaid/control/nat/r.py , and given the size of the file it's a nontrivial amount of work, but would be nice to have nevertheless.