# is forward compatible beyond the current compute capability major
# release while SASS is only forward compatible inside the current
# major release. Example: sm_80 kernels can run on sm_89 GPUs but
# not on sm_90 GPUs. compute_80 kernels though can also run on sm_90 GPUs.
common:cuda_clang --repo_env=HERMETIC_CUDA_COMPUTE_CAPABILITIES="sm_60,sm_70,sm_80,sm_89,compute_90"
# Permit newer CUDA versions than Clang is aware of

* Fix logrotate config on GCI ([#29139](https://github.com/kubernetes/kubernetes/pull/29139), [@adityakali](https://github.com/adityakali))
* Do not query the metadata server to find out if running on GCE.  Retry metadata server query for gcr if running on gce. ([#28871](https://github.com/kubernetes/kubernetes/pull/28871), [@vishh](https://github.com/vishh))

turn"4.6.2"}},{key:"Default",get:function(){return at}},{key:"DefaultType",get:function(){return st}}]),t}();o.default(document).on(nt,it,lt._dataApiKeydownHandler).on(nt,ot,lt._dataApiKeydownHandler).on(et+" keyup.bs.dropdown.data-api",lt._clearMenus).on(et,it,(function(t){t.preventDefault(),t.stopPropagation(),lt._jQueryInterface.call(o.default(this),"toggle")})).on(et,".dropdown form",(function(t){t.stopPropagation()})),o.default.fn[z]=lt._jQueryInterface,o.default.fn[z].Constructor=lt,o.defa...

routing_key    (string)    routing key for publishing
mandatory      (on|off)    quietly ignore undelivered messages when set to 'off', default is 'on'
durable        (on|off)    persist queue across broker restarts when set to 'on', default is 'off'
no_wait        (on|off)    non-blocking message delivery when set to 'on', default is 'off'

_2020-04-28_

 *  Fix: Don't crash on Java 8u252 which introduces an API previously found only on Java 9 and
    above. See [Jetty's overview][jetty_8_252] of the API change and its consequences.


## Version 3.12.10

_2020-02-29_

 *  Fix: Don't crash on Android 4.1 when detecting methods that became restricted in Android 11.
    Supporting a full decade of Android releases on our 3.12.x branch is tricky!

// `run_metadata` (when not NULL) and should manually call TF_DeleteBuffer on
// them.
//
// On success, the tensors corresponding to outputs[0,noutputs-1] are placed in
// output_values[]. Ownership of the elements of output_values[] is transferred
// to the caller, which must eventually call TF_DeleteTensor on them.
//
// On failure, output_values[] contains NULLs.
TF_CAPI_EXPORT extern void TF_SessionRun(

   * There is no need to use this method on a {@link Map} implementation that already has a
   * corresponding {@link Set} implementation (such as {@link java.util.HashMap} or {@link
   * java.util.TreeMap}).
   *
   * <p>Each method invocation on the set returned by this method results in exactly one method
   * invocation on the backing map or its {@code keySet} view, with one exception. The {@code

	// Reorder online disks based on erasure distribution order.
	// Reorder parts metadata based on erasure distribution order.
	onlineDisks, metaArr = shuffleDisksAndPartsMetadataByIndex(onlineDisks, metaArr, fi)

	// For negative length read everything.
	if length < 0 {

   * noted, but the initial table is a zero-length array.
   *
   * <p>This test may fail miserably on non-OpenJDK environments...
   */
  @J2ktIncompatible
  @GwtIncompatible // reflection
  @AndroidIncompatible // relies on assumptions about OpenJDK
  public void testNewHashMapWithExpectedSize_wontGrow() throws Exception {
    // before jdk7u40: creates one-bucket table

  }

  /**
   * Like {@link #transform(ListenableFuture, Function, Executor)} except that the transformation
   * {@code function} is invoked on each call to {@link Future#get() get()} on the returned future.
   *
   * <p>The returned {@code Future} reflects the input's cancellation state directly, and any
   * attempt to cancel the returned Future is likewise passed through to the input Future.