Aaron Lun

I'm not quite sure why I did this, but the current code at https://github.com/Bioconductor/SummarizedExperiment/blob/8df97720354bdeecaf947f574dddd09da02d55b2/R/combine-methods.R#L244 wipes out the names of the metadata list, e.g., ```r example(SummarizedExperiment, verbose=FALSE) metadata(se)

Consider the following example: ```r library(SummarizedExperiment) se

derived RSE to SE coercion discards names

15

Coercing a subclass of an RSE to an SE discards the row names: ```r library(SummarizedExperiment) example(SummarizedExperiment, echo=FALSE) setClass("MyExperiment", contains="RangedSummarizedExperiment") me

Inspired by https://support.bioconductor.org/p/126722/. The problem is what to do when a `rowData` field is not identical across `cbind`'d elements. Currently `cbind` throws an error, which is understandable (and safest) but...

colProds and rowProds should accept method= argument

8

Even if it doesn't do anything with them. Just add a `...` to both method definitions so that they don't throw when someone passes in `method=` in a generic.

Just a thought, but would be interesting to think about whether we could have: - A standalone C/C++ library for **zarr**. - Something like **Rhdf5lib**, but wrapping the above library....

DelayedMatrixStats no longer supports scalar `center=` arguments

5

PeteHaitch/DelayedMatrixStats@9048f3e7a24035f3526a2ac0455598138e0c8480 causes https://github.com/Bioconductor/DelayedArray/blob/700cfdc38e0f97a67942963c1555d0fd83a7bc41/R/DelayedArray-utils.R#L868 to fail, for example: ```r library(DelayedArray) mat

More informative error than `the supplied seed must support extract_array()`

3

A recent change in the DelayedArray stack (don't know exactly where) is causing **ScaledMatrix** tests to [break on Bioc-devel](https://master.bioconductor.org/checkResults/3.19/bioc-LATEST/ScaledMatrix/nebbiolo1-checksrc.html) with the rather unhelpful message: ```r Error in `validObject(.Object)`: invalid class...

Thoughts on migrating from `read_sparse_block()` to SVTs

2

Currently, `read_sparse_block()` returns the old `SparseArraySeed`, so I figured I would try updating some of my packages to the new SVT world. It seems `SparseArray::extract_sparse_array()` or `SparseArray::read_block_as_sparse()` are the replacements,...

I could have sworn this used to work. ```r library(DelayedArray) x