caused an IOException.
 * Fix bug in HTTP/2 where our HPACK decoder wasn't emitting entries in
   certain eviction scenarios, leading to dropped response headers.

## Version 1.5.1

_2014-03-11_

 * Fix 1.5.0 regression where connections should not have been recycled.
 * Fix 1.5.0 regression where transparent Gzip was broken by attempting to
   recover from another I/O failure.

    could use new flow control capacity before acknowledging it, causing strict HTTP/2 servers to
    fail the call.

 *  Fix: Recover gracefully when a coalesced connection immediately goes unhealthy.

## Version 3.14.2

_2019-05-19_

 *  Fix: Lock in a route when recovering from an HTTP/2 connection error. We had a bug where two

# OpenAPI Webhooks

There are cases where you want to tell your API **users** that your app could call *their* app (sending a request) with some data, normally to **notify** of some type of **event**.

This means that instead of the normal process of your users sending requests to your API, it's **your API** (or your app) that could **send requests to their system** (to their API, their app).

This is normally called a **webhook**.

## Webhooks steps

    including a bug where the cache could be corrupted upon certain access
    patterns. We also fixed a bug where the cache was being cleared due to a
    corrupted journal. We've added APIs to configure a request's `Cache-Control`
    headers, and to manually clear the cache.

 *  **Request cancellation fixes.** This update fixes a bug where synchronous

     permission under the Copyright and Similar Rights held by the
     Licensor. For purposes of this Public License, where the Licensed
     Material is a musical work, performance, or sound recording,
     Adapted Material is always produced where the Licensed Material is
     synched in timed relation with a moving image.

  b. Adapter's License means the license You apply to Your Copyright

# Where component owners are known, add them here.

/tensorflow/c/eager @qqfish
/tensorflow/core/common_runtime/eager @qqfish
/tenosrflow/core/debug @caisq
/tensorflow/core/kernels/mkl/ @penpornk
/tensorflow/core/kernels/sparse/ @penpornk
/tensorflow/core/nccl/ @azaks2 @chsigg
/tensorflow/python/autograph/ @mdanatg
/tensorflow/python/debug @caisq
/tensorflow/python/eager @rohan100jain
/tensorflow/tools/docs/ @markdaoust

* Implement and use well-known cryptographic tools, like Passlib and JWT tokens, etc.
* Add more granular permission controls with OAuth2 scopes where needed.
* ...etc.

Nevertheless, you might have a very specific use case where you really need to disable the API docs for some environment (e.g. for production) or depending on configurations from environment variables.

## Conditional OpenAPI from settings and env vars

# ==============================================================================
#
# Changes the behavior in pycpp.sh from "run all tests" to "verify that all
# tests can compile." Used in some CI jobs (macOS and Linux Arm64) where test
# execution is too expensive.
TFCI_PYCPP_SWAP_TO_BUILD_ENABLE=1

 *  Fix: Don't crash decompressing web sockets messages. We had a bug where we assumed deflated
    bytes in would always yield deflated bytes out and this isn't always the case!

 *  Fix: Reliably update and invalidate the disk cache on windows. As originally designed our
    internal `DiskLruCache` assumes an inode-like file system, where it's fine to delete files that

       pulling in other changes.
 - [ ] Each commit in the pull request should have a meaningful subject line and body.
 - [ ] Format the pull request title like `[MNG-XXX] SUMMARY`,
       where you replace `MNG-XXX` and `SUMMARY` with the appropriate JIRA issue.
 - [ ] Also format the first line of the commit message like `[MNG-XXX] SUMMARY`.