Lee Hambley

This is too big of a change to test. Others work around this by injecting a module into the backend class. Others by passing the execution context into their own...

If I'm reading this correctly the bigger (and recurring) problem is that people are surprised by the internal use of `instance_exec` and surprised that the variable scope changes. I dare...

The issue with `instance_exec` is one of my most regretted design decisions with SSHKit, as I don't really understand the issue and how it specifically affects library authors, could someone...

> AFAIK nested on blocks were never considered in the design. The fact that it works at all is really just coincidence, due to how the SSHKit DSL is mixed...

> docs aren't good enough to explain the design choice IMHO - just check > stackoverflow... Alright, that's a problem. Here's the thing, we need a list of commands that...

What cases can you offer up as sane use-cases for strings in commands, and can we take those and see if we can make the whole API work reliably?

Alright, so thanks for those suggestions, better than I've come across, whilst I could shoot them all down with "you should be doing X" (and, maybe that's the resolution we...

> i wonder if generating a _script_ per command to be run would be better than > jamming it all into a bash/sh cli Sounds like a rather large departure...

Because the `test` contains a space https://github.com/leehambley/sshkit/blob/master/EXAMPLES.md#run-a-command-without-it-being-command-mapped

For test() the use-case is a bit unusual, granted… I didn't come across any cases where it felt too bad, but if you have a better idea than "if it...