"in": "query",
            "name": "continue",
            "schema": {
              "type": "string",
              "uniqueItems": true
            }
          },

        else -> Socket(route.proxy)
      }
    this.rawSocket = rawSocket

    // Handle the race where cancel() precedes connectSocket(). We don't want to miss a cancel.
    if (canceled) {
      throw IOException("canceled")
    }

    rawSocket.soTimeout = socketReadTimeoutMillis
    try {

		// parse errors will be built (and fail) instead of being silently skipped
		// as not matching the pattern. Go 1.5 and earlier skipped, but that
		// behavior means people miss serious mistakes.
		// See golang.org/issue/11407.
		if p, err := cfg.BuildContext.ImportDir(path, 0); err != nil && (p == nil || len(p.InvalidGoFiles) == 0) {
			if _, noGo := err.(*build.NoGoError); noGo {

				"About to decode encrypted object",
				"Decoded encrypted object",
				"About to decrypt data using DEK",
				"Data decryption succeeded",
			},
		},
		{
			desc:     "decrypt with cache miss",
			cacheTTL: 1 * time.Second,
			expected: []string{
				"About to decode encrypted object",
				"Decoded encrypted object",
				"About to decrypt DEK using remote service",
				"DEK decryption succeeded",

A task can only be loaded from the cache if it has stable task inputs.
Unstable task inputs result in the task having a unique set of inputs for every build, which will always result in a cache miss.

[[normalization]]
== Better reuse via input normalization

Having stable inputs is crucial for cacheable tasks.
However, achieving byte for byte identical inputs for each task can be challenging.

            "in": "query",
            "name": "continue",
            "schema": {
              "type": "string",
              "uniqueItems": true
            }
          },

          "in": "query",
          "name": "continue",
          "schema": {
            "type": "string",
            "uniqueItems": true
          }
        },
        {

// event generator yet.
//
// Note that GenericPLEG assumes that a container would not be created,
// terminated, and garbage collected within one relist period. If such an
// incident happens, GenenricPLEG would miss all events regarding this
// container. In the case of relisting failure, the window may become longer.
// Note that this assumption is not unique -- many kubelet internal components

          "in": "query",
          "name": "continue",
          "schema": {
            "type": "string",
            "uniqueItems": true
          }
        },
        {

          "in": "query",
          "name": "continue",
          "schema": {
            "type": "string",
            "uniqueItems": true
          }
        },
        {