Morten Piibeleht

Hmm, this particular error will pop up when you just start `julia` and you have the `atreplinit()` logic in `startup.jl`. Would it be possible to have the `@async Revise.wait_steal_repl_backend()` line...

Hmm, I indeed didn't look at the code that closely. However, if I understand correctly, `steal_repl_backend` does not get called in `__init__` unless you are doing `using Revise` in the...

Also, with this patch, I think the recommended way to load Revise in the `startup.jl` file would become just `using Revise`. If I have this ```julia atreplinit() do repl try...

Ah, the warning showed up. It was the same one that @mlhetland mentioned in https://github.com/timholy/Revise.jl/issues/202#issuecomment-518343458. I had `using Revise` in `startup.jl` and there was some compilation necessary, so this happened:...

Ok, I've tested this wrapped version and it seems to work great. The warning is gone and loading Revise in either `startup.jl` or `-e/E` still works. Are there any tests...

Would it make sense for the logic to be "Try Cholesky. If error, try QR"? I.e so that the user wouldn't have to do anything?

#1747 contains some unnecessary changes as far as I can tell. But I think the general approach there is good (create a new expander step that recurses for certain elements,...

You'd have to start a new PR on your fork, but you can use their latest commit as a starting point (although, it might be easier to just re-do it...

Not 100% sure if it's related, but this ```julia function myfunction_foo(io::IO) write(io, """ """) end ``` goes into ```julia function myfunction_foo(io::IO) write( io, """ """, ) end ``` which does...

That doesn't seem necessary? The string wouldn't change? So it could just go into ```julia function myfunction_foo(io::IO) write( io, """ """, ) end ``` Or if we do insist on...