In contrast, the fix uses the outermost function name from sys.calls(), which corresponds with my wrapper function.
It then tries to copy the non-existent value into "AzimuthReference", which deletes the desired entry.
I think this can be fixed by using the innermost (last) function name from sys.calls(), as this is the function that calls Tool<-.
Hey - I recently started using Azimuth and have had some trouble similar to #155 when building a reference:
It seems the fix (b1b6895) has a bug in the logic trying to generate the automatically assigned tool name.
The
Tool<-
method gets the function name that calledTool<-
, per https://github.com/satijalab/seurat-object/blob/release/5.0.1/R/seurat.R#L2636. In this case, theTool<-
call uses the correct "AzimuthReference".In contrast, the fix uses the outermost function name from
sys.calls()
, which corresponds with my wrapper function. It then tries to copy the non-existent value into "AzimuthReference", which deletes the desired entry.I think this can be fixed by using the innermost (last) function name from
sys.calls()
, as this is the function that callsTool<-
.For reference, I am running https://github.com/satijalab/azimuth/tree/release/0.5.0.