Use subcommands rather than flags about delve

[pnasrat]

It depends on who your audience is - my experience with docker is some users have a dislike of go style flags as non-getopty/traditional.

The command based approach would be more similar to the go tool itself, and subcommands could have flags. I think I've definitely feel -run being a boolean doesn't make much sense.

dlv --run=false --pid 10027
dlv --run=true --pid 10027

dlv attach -p

Note then attach could also work for a host:port

Personally whatever works for invocation - but I think I'd tend to lean towards subcommands and short flags or args. The use case I can think of are:

• Load a program (with args) without executing - enter delve interface, run delve commands (set breakpoints, etc then execute within delve (either single step or just run/cont)
• Load a program (with args) without executing and a core file for postmortem debugging
• Attach to a running program (local)
• Attach to a running program (remote)
• Run with a program/attached to a program with scripted input (batch mode)
• Build and load