Logging enhancements

[shaunn]

Background: When making changes that may affect multiple components/clusters having the generate fail deadly has become personally an issue when attempting to get an overview of what problems exist past the one that fatally exited.

Two arguments --noexit and --long are introduced; the first to stop the program from exiting during a fatal or panic event, and the second to print error messages in a multi-line readable format.

Build params:

• go version go1.19.6 darwin/amd64

Example of how a run with a fatal error works, just as before

A generate with the --long flag showing multi-line print of the error (fmt.Println(err)) for the humans before immediately exiting

A generate with the --noexit flag demonstrating that fatal errors do not exit and all runs attempt to continue

A generate with the --noexit and --long flags together demonstrating that fatal errors do not exit and all runs attempt to continue AND all errors printed in human readable format

Updated --long

[cspargo-apptio]

zerolog already has a facility for this to append custom output

        log.Logger = log.Output(zerolog.ConsoleWriter{Out: os.Stderr, NoColor: !colorOutput,
                FormatExtra: func(evt map[string]interface{}, buf *bytes.Buffer) error {
                        if errstring, ok := evt["error"]; ok {
                                buf.WriteString("\n" + fmt.Sprint(errstring))
                        }
                        return nil
                },
        })

[shaunn]

@cspargo-apptio Yes, thank you! I was tripped up on this. Works perfectly!

[srosenberg-apptio]

I would close this if I could