Missing inward/outward mechanism on TCP / HTTP remote act

[jeromevalentin]

Seneca 3.2 introduced the mechanism of inward/outward on act allowing to intercept action before/after their executions. This allows to upgrade act input / output dynamically on the fly.

The same kind of mechanism is missing for remote acts.

Exemple:

    CreateInstance()
      .add('foo:1', function (args, done) {
        done(null, { BAR: args.bar })
      })
      .listen({type: 'http', port: '18997',
        inward: (context, { msg }) => {
          msg.bar += 1
          msg.inward = 'INPUT UPGRADED'
        },
        outward: (context, { msg, res, err }) => {
          res.BAR += 10
          res.inward = msg.inward
          res.outward = 'OUTPUT UPGRADED'
        }
      })
      .ready(function () {
        var siClient = CreateInstance()
          .client({type: 'http', port: '18997'})

        siClient.act('foo:1,bar:2', function (err, out) {
          if (err) return fin(err)
          Assert.equal(out.BAR, 13)
          Assert.equal(out.inward, 'INPUT UPGRADED')
          Assert.equal(out.outward, 'OUTPUT UPGRADED')
          fin()
        })
      })

Here the proposal is to use the listen options to register the inward and outward hooks.

Such mechanism would allow to:

• log remote acts easily
• perform some security checking (like user authentication) in inward before really performing the act with the option to reject the act
• perform transparently some clean up on remote act results (like removing all $ properties on exposing entities for example)