Merge runs forever without error if x is inside a named vector

[edridgedsouza]

Had a part of my script that looked like this:

all_objs <- c()
for (i in seq_along(dirs)) {
    samplename <- basename(dirs[i])
    obj <- Load10X_Spatial(data.dir=dirs[i], 
                            filename='filtered_feature_bc_matrix.h5')
    all_objs[[samplename]] <- obj
}

obj_combined <- merge(all_objs[1], y = all_objs[-1], 
                      add.cell.ids = names(all_objs), 
                      project = "Spatial Seq")

This was taking upwards of several hours, even when I restricted my all_objs to contain only two objects. After some tinkering, I found that the offending part was inside the merge() operation: the argument x here is technically a Seurat object inside a vector, rather than the object itself. Changing x to all_objs[[1]], without making any changes to y, seemed to do the trick and the samples merged in a few seconds.

Ultimately, this is on me for not reading the docs more closely to realize that x should literally be the object itself, and not the object contained inside a named vector (which isn't the case for y). However, I can imagine that at least a few other people may accidentally make such a mistake and not realize anything is wrong for a while; my htop showed 100% CPU usage so I assumed the function was running normally and was simply a lot more computationally intensive than I thought.

No idea why this bug occurs, since merge.Assay in assay.R simply performs c(x, y) under the hood anyways. If not fixable, would it be possible to add a warning message?

Relevant package versions: SeuratObject_4.1.3 Seurat_4.3.0 spacexr_2.2.1