search

SydneyBioX / MoleculeExperiment

https://sydneybiox.github.io/MoleculeExperiment/
Other
12 stars 0 forks source link

Please take care in using sp #1

Closed rsbivand closed 1 year ago

rsbivand commented 1 year ago

Please read all of https://r-spatial.org/r/2023/05/15/evolution4.html and the previous reports. The sp package is really for geographical data, and has largely been superceded by sf (S3) and terra (S4 but only pointers to objects in memory). My guess is that if "spatial" cluster packages in BioConductor need classes, you'd need either to borrow from what geo-spatial CRAN packages are doing, or look to start an R7 process to succeed sp. https://github.com/SydneyBioX/MoleculeExperiment/blob/8af3918aaa8b5031ab94a14f33e0799aabc72bb7/R/countMolecules.R#L243 looks very vulnerable, as sp has used rgeos(to be archived in October 2023) for some over methods. Your NAMESPACE file does not import anything from sp.

shazanfar commented 1 year ago

Hi @rsbivand , thank you for bringing these upcoming changes to sp to our attention. I'm pleased that we have already implemented a new faster function based on terra, and the sp dependency belongs to the function that has already been superceded - we had kept this for internal development and checking. Therefore, we have been able to straightforwardly remove the dependency on sp by just removing the function in question completely, and just use terra. Thank you!