consider exposing Readline.completion_proc via HighLine::Question#selection

[doriantaylor]

consider exposing Readline.completion_proc via HighLine::Question#selection

[abinoam]

Hi @doriantaylor , thank you for your comments/suggestions.

I'll probably have time to look at it further by the weekend.

But I can antecipate to you that if it solves an actual problem that you're having, it doesn't break the api and it doesn't raise the code's complexity too much I'll probably welcome your changes 👍

[doriantaylor]

Great, I'll cook it up.

[doriantaylor]

Looking over this and I have a couple questions:

1. Testing strategy: Do I (can I?) use HighLine::Simulate or do I add another manual acceptance test?
2. The method HighLine#shell_style_lambda does something similar to what I am ultimately trying to do, but only for menus. Perhaps it would make sense to take a step back and look at this more holistically?

I found my way here because I was writing a command-line tool using commander and wanted to implement an umbrella shell command that recycled the other available commands into a readline loop. (To achieve this I had to patch some undesirable behaviour, which the maintainer graciously applied.) Commander, of course, uses HighLine for its readline component, and I want to be able to add tab completion—not just for the commands, but for the options as well.

Given that my desired functionality is at least partially present in HighLine::Menu, I should probably take it into consideration. Nevertheless, I believe it to be still overall useful to be able to access Readline.completion_proc directly, it just looks like a slightly more delicate surgery.