Supporting non-JS debuggers

[hediet]

Supporting non-JS debuggers

[hediet]

That's a pretty good idea! However, it might reduce the incentive to add proper support for all the specific debug adapters (which I think is fine).

I'm not a C++ guy and lldb/gdb are rather alien to me. I'm very curious in which ways these evalTemplates would be used and what could be achieved by them. It would be awesome if these use-cases or setups could be collected in this repository.

For what do you need the differentiation between watch or repl contexts though? The extension always uses the watch context, as the repl context sometimes is more verbose, making the result unparsable.

Oh, by the way, thanks for your great lldb vscode extension ;)

[vadimcn]

it might reduce the incentive to add proper support for all the specific debug adapters (which I think is fine).

If templates are configurable, others will be able to create them and contribute back to your repo. Otherwise, in order to add a new debugger, one would have to fork and build your extension, which raises the bar considerably.

For what do you need the differentiation between watch or repl contexts though?

In the case of lldb extension, debugger commands may be executed only in "repl" context. Other debuggers may have this as well, so IMO, it'd be best to let developers choose the context.

[hediet]

There is now the configuration option debugVisualizer.debugAdapterConfigurations in 0.13.0:

"debugVisualizer.debugAdapterConfigurations": {
	"type": "object",
	"additionalProperties": {
		"type": "object",
		"properties": {
			"expressionTemplate": {
				"type:": "string",
				"description": "A template for the final expression sent to the debugger for evaluation. Use `${expr}` to refer to the currently visualized expression."
			},
			"context": {
				"type": "string",
				"enum": [
					"watch",
					"repl"
				],
				"description": "The context to use for evaluating the expression."
			}
		}
	}
}

[hediet]

Closing as this should solve your use case.

[hediet]

Hey @vadimcn! Which context do you think should be used by default for evaluating expressions?
I used to chose watch as default, but many debug adapters truncate long strings, so I switched to repl. But for repl, the print command of lldb wraps the result in some other text. Also, having to use the print command is confusing.