@joaomoreno welcome to the party: #58116 (comment)

The type of options should be extracted to an independent type: OutputChannelOptions

@sandy081 I actually don't know why we have options instead of a simple boolean?

inMemory was actually not too bad! Others:

• transient
• noPersist
• persistent (defaults to true)
• useLogFile (defaults to true)

Since it's not very intuitive we try to avoid something optional that defaults to true

@jrieken I think you had the same question during review and it is because

There is a request to set a lang id for channels. If so it has to be set while creating. Hence created options object.

I also think inMemory as a better option even though we use log files on the renderer side when true. When channel is open, data directly goes into in memory model where as in other case it first goes to file and then to the model.

other case it first goes to file and then to the model.

and that means it is not in memory?

Eventually, for all channels, content gets loaded in to model which is in memory to show in the UI. But the difference is that those channels with special flag, directly transmit data to renderer from ext host, in which case data remains in memory.

Well, you decide, these are the constraints

• don't have something optional that default to true
• because we want people to use the log-file use a name that kinda means 'I am doing something bad/expensive'
• don't use a technical term, esp. not inMemory because that also shows up as scheme

Last, be aware that you created this problem just for yourself (trying to avoid a third argument in the future) and that using a parameter gives you the flexibility to change it later

/**
 * Options to configure the behavior of the [OutputChannel](#OutputChannel)
 */
export interface OutputChannelOptions {

	/**
	 * By default, data appended to Output channels is written into a log file and is not piped to UI process immediately.
	 * Output channel UI is updated by listening to the corresponding log file changes.
	 * This helps to reduce CPU load on the UI process but also means that the output channel UI updates with a small delay.
	 *
	 * Enabling this flag will allow the data to be piped to UI process for immediate updates at the cost of higher CPU usage.
	 */
	readonly enablePiping?: boolean

}

/**
 * Creates a new [output channel](#OutputChannel) with the given name.
 *
 * @param name Human-readable string which will be used to represent the channel in the UI.
 * @param options Options to configure the behavior of the channel.
 */
export function createOutputChannel(name: string, options?: OutputChannelOptions): OutputChannel;

What do you think?

What do you think?

I think you should make it a param instead of an object and pick any name because that's not set in stone

Removed the options completely.