excluding those notices that do not pertain to any part of
          the Derivative Works; and

      (d) If the Work includes a "NOTICE" text file as part of its
          distribution, then any Derivative Works that You distribute must
          include a readable copy of the attribution notices contained
          within such NOTICE file, excluding those notices that do not

	c_ISUID = 04000 // Set uid
	c_ISGID = 02000 // Set gid
	c_ISVTX = 01000 // Save text (sticky bit)

	// Common Unix mode constants; these are not defined in any common tar standard.
	// Header.FileInfo understands these, but FileInfoHeader will never produce these.
	c_ISDIR  = 040000  // Directory
	c_ISFIFO = 010000  // FIFO
	c_ISREG  = 0100000 // Regular file
	c_ISLNK  = 0120000 // Symbolic link

)

// tagKey returns the tag-based key to use in the pkgCache.
// It is a comma-separated string; the first part is dir, the rest tags.
// The satisfied tags are derived from context but only those that
// matter (the ones listed in the tags argument plus GOOS and GOARCH) are used.
// The tags list, which came from go/build's Package.AllTags,
// is known to be sorted.

# certain shared libraries during the wheel building process (https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/pip_package/build_pip_package.sh#L255-L262).
# When we use Patchelf versions <0.12, those shared libraries end up with a
# corrupted PT_NOTE program header. This was fixed in v0.12, see https://github.com/NixOS/patchelf/commit/43a33482b501b0f5ee9da312aabfca3806570cc9.

    -   Uses `pycpp.sh`
-   Presubmit jobs (Run on every GitHub PR)
    -   Uses `pycpp.sh`, `code_check_changed_files.sh`

These "env" files match up with an environment matrix that roughly covers:

-   Different Python versions
-   Linux, MacOS, and Windows machines (these pool definitions are internal)
-   x86 and arm64
-   CPU-only, or with NVIDIA CUDA support (Linux only), or with TPUs

process ends, those variables are part of one or more initialization cycles,
and the program is not valid.
</p>

<p>
Multiple variables on the left-hand side of a variable declaration initialized
by single (multi-valued) expression on the right-hand side are initialized
together: If any of the variables on the left-hand side is initialized, all
those variables are initialized in the same step.
</p>

***NOTE***: Only original source code from you and other people that have signed the CLA can be accepted into the main repository.

### Contributing code

If you have improvements to TensorFlow, send us your pull requests! For those
just getting started, GitHub has a
[how-to](https://help.github.com/articles/using-pull-requests/).

TensorFlow team members will be assigned to review your pull requests. Once the

Programs that modify data being simultaneously accessed by multiple goroutines
must serialize such access.
</p>

<p>
To serialize access, protect the data with channel operations or other synchronization primitives
such as those in the <a href="/pkg/sync/"><code>sync</code></a>
and <a href="/pkg/sync/atomic/"><code>sync/atomic</code></a> packages.
</p>

<p>
If you must read the rest of this document to understand the behavior of your program,

	//
	// dataDescriptorLen includes the size of the signature but
	// first read just those 4 bytes to see if it exists.
	if _, err := io.ReadFull(r, buf[:4]); err != nil {
		return err
	}
	off := 0
	maybeSig := readBuf(buf[:4])
	if maybeSig.uint32() != dataDescriptorSignature {
		// No data descriptor signature. Keep these four
		// bytes.
		off += 4
	}
	if _, err := io.ReadFull(r, buf[off:12]); err != nil {

		// local character encoding. Most encoding are compatible with a large
		// subset of CP-437, which itself is ASCII-like.
		//
		// Forbid 0x7e and 0x5c since EUC-KR and Shift-JIS replace those
		// characters with localized currency and overline characters.
		if r < 0x20 || r > 0x7d || r == 0x5c {
			if !utf8.ValidRune(r) || (r == utf8.RuneError && size == 1) {
				return false, false
			}