Closed rcannood closed 3 months ago
Good idea! I was going to add it to the roxygen docs, but found it a bit weird to document differences between R and Python while the Python AnnData also only allows string obs_names and var_names for now. I'll create a separate issue.
I have no issues with the way you refactored
obs_names
andvar_names
.If I recall correctly though, the reason for not doing it this way in the first place was the following: rownames in R can only be character vectors, which is an assumption that does not (or will not) apply to the Python anndata version. I believe that the impact of this is relatively minor, however I do feel it is necessary to be documented somewhere? (This is in general: we need to have some documentation delineating the differences and what is and isn't supported)
You are right. I had forgotten but this was (at least partly) why we changed it before. I would probably make a special man page and/or vignette for things we know are different from Python.
I would probably make a special man page and/or vignette for things we know are different from Python.
Good point! I added it to #172 :relaxed:
These changes have been ported from #166 and #169.
Rownames have been moved from a separate obs_names and var_names to being the rownames of obs and var, since it simplifies the way the data is read to and from an hdf5 file.