Daniel Karrasch

Well, as expected, if I introduce ```julia function Base.:(*)(A1::LinearMap, A2::UniformScaling{Bool}) if A2.λ return A1 else return A1 * A2.λ # this could be further reduced to a new ZeroMap type...

I've been thinking about it. The issue is that this introduces type instability. If we restrict this to only `Bool`, then the effect is somewhat limited. OTOH, maybe the product...

I've been thinking and working on that issue here now for a while. It seems to me, that this is probably not an issue with `LinearMaps.jl`. Both attempts to solve...

Many things have been said already, so let me add my two cents here. First, making `LinearMap` subtype `AbstractMatrix` is a one-line change, which immediately reveals at least one major...

@JeffFessler I suspect you are using `FunctionMap`s heavily. Can you please try your packages with this PR branch and report back any new issues? I believe nothing should break, only...

Sorry I messed a few things up while splitting. Seems fixed now.

Thank you very much for the detailed report! If you received deprecation warnings (no matter how many), then it grabbed the right branch. I'm glad that tests pass, too. I...

Drive-by comments are always welcome! There is dispatch on `IIPFunctionMap` vs. `OOPFunctionMap` in linearcombinations.jl and composed.jl. And yes, it's used to catch cases where all `FunctionMap`s allocate the result array...

Hm, another option could be to make `LinearMaps.ismutating` a "trait" for all `LinearMap` subtypes. That would allow asking a `LinearCombination`/`CompositeMap` whether all its maps are mutating, whereas the current design...

I realized we already have some kind of the necessary machinery: `MulStyle`, which so far has only `ThreeArg` and `FiveArg`, but which could be extended to `TwoArg`, meaning `A*x`. Reflecting...