* higher value than you need can waste space and time, and a significantly lower value can lead
   * to thread contention. But overestimates and underestimates within an order of magnitude do not
   * usually have much noticeable impact. A value of one permits only one thread to modify the cache
   * at a time, but since read operations and cache loading computations can proceed concurrently,

config_setting(
    name = "no_xla_deps_in_cuda",
    define_values = {"no_xla_deps_in_cuda": "true"},
    visibility = ["//visibility:public"],
)

# Crosses between framework_shared_object and a bunch of other configurations
# due to limitations in nested select() statements.
config_setting(
    name = "framework_shared_object",
    define_values = {"framework_shared_object": "true"},

   * higher value than you need can waste space and time, and a significantly lower value can lead
   * to thread contention. But overestimates and underestimates within an order of magnitude do not
   * usually have much noticeable impact. A value of one permits only one thread to modify the cache
   * at a time, but since read operations and cache loading computations can proceed concurrently,

   *
   * <p><i>Performance notes:</i> while the cartesian product of sets of size {@code m, n, p} is a
   * set of size {@code m x n x p}, its actual memory consumption is much smaller. When the
   * cartesian set is constructed, the input sets are merely copied. Only as the resulting set is
   * iterated are the individual lists created, and these are not retained after iteration.
   *

	if x == o {
		return false
	}
	// Prefer newest modtime.
	if x.ModTime != o.ModTime {
		return x.ModTime > o.ModTime
	}

	// The following doesn't make too much sense, but we want sort to be consistent nonetheless.
	// Prefer lower types
	if x.Type != o.Type {
		return x.Type < o.Type
	}
	// Consistent sort on signature

    timeToFirstByte()
  }

  /**
   * Test to confirm that events are reported at the time they occur and no earlier and no later.
   * This inserts a bunch of synthetic 250 ms delays into both client and server and confirms that
   * the same delays make it back into the events.
   *

   *
   * <p><i>Performance notes:</i> while the cartesian product of sets of size {@code m, n, p} is a
   * set of size {@code m x n x p}, its actual memory consumption is much smaller. When the
   * cartesian set is constructed, the input sets are merely copied. Only as the resulting set is
   * iterated are the individual lists created, and these are not retained after iteration.
   *

# Do not do this. If enabled protobuf's core internal target 
# @com_google_protobuf//python:protobuf_python will start depending on a bunch
# of cc_binary shared libraries artifacts, which will mess with how we link 
# protobuf dependencies ourselves. By default this value is false, but some 
# projects enable it, which we don't want here.

   *
   * if (guardExpression) {
   *    throw new BadException(badMsg(...));
   * }
   *
   * The alternative natural refactorings into void or Exception-returning methods are much slower.
   * This is a big deal - we're talking factors of 2-8 in microbenchmarks, not just 10-20%. (This is
   * a hotspot optimizer bug, which should be fixed, but that's a separate, big project).
   *

   *
   * if (guardExpression) {
   *    throw new BadException(badMsg(...));
   * }
   *
   * The alternative natural refactorings into void or Exception-returning methods are much slower.
   * This is a big deal - we're talking factors of 2-8 in microbenchmarks, not just 10-20%. (This is
   * a hotspot optimizer bug, which should be fixed, but that's a separate, big project).
   *