Summation of ConstantArray of 0 with vector of 0's fails

Reproduce the zero-valued case starting with DelayedArray::ConstantArray and the HDF5Array::writeHDF5Array entry point, then compare it with the working combinations shown in the report. Done means the zero ConstantArray plus zero vector can be written successfully, with the existing examples serving as regression checks.

Hi,

I ran into a curious edge case, where summing a ConstantArray with a vector fails if both are zero and then writing it with HDF5Array::writeHDF5Array, but not in any other combination:

mat <- DelayedArray::ConstantArray(c(2, 4), value = 0)
res <- mat + rep(0, 2)
res
#> <2 x 4> sparse DelayedMatrix object of type "double":
#>      [,1] [,2] [,3] [,4]
#> [1,]    0    0    0    0
#> [2,]    0    0    0    0
HDF5Array::writeHDF5Array(res)
#> Error in .Arith_SVT1_v2(.Generic, e1, e2): "+" is not supported between a SparseArray object and a numeric vector
#>   (result wouldn't be sparse in general)

# All other combination seems to work
mat <- DelayedArray::ConstantArray(c(2, 4), value = 0)
res <- mat + rep(3, 2)
HDF5Array::writeHDF5Array(res)
#> <2 x 4> HDF5Matrix object of type "double":
#>      [,1] [,2] [,3] [,4]
#> [1,]    3    3    3    3
#> [2,]    3    3    3    3

mat <- DelayedArray::ConstantArray(c(2, 4), value = 17)
res <- mat + rep(0, 2)
HDF5Array::writeHDF5Array(res)
#> <2 x 4> HDF5Matrix object of type "double":
#>      [,1] [,2] [,3] [,4]
#> [1,]   17   17   17   17
#> [2,]   17   17   17   17

mat <- DelayedArray::ConstantArray(c(2, 4), value = 0)
res <- mat + 0
HDF5Array::writeHDF5Array(res)
#> <2 x 4> sparse HDF5Matrix object of type "double":
#>      [,1] [,2] [,3] [,4]
#> [1,]    0    0    0    0
#> [2,]    0    0    0    0

^{Created on 2025-03-05 with reprex v2.1.1}

I am not actually 100% this is an DelayedArray issue or if it might rather need to be moved to HDF5Array.

Best,
Constantin