## Version 2.2.0

_2014-12-30_

 *  **`RequestBody.contentLength()` now throws `IOException`.**
    This is a source-incompatible change. If you have code that calls
    `RequestBody.contentLength()`, your compile will break with this
    update. The change is binary-compatible, however: code compiled
    for OkHttp 2.0 and 2.1 will continue to work with this update.

Et comme on peut avoir du parallélisme et de l'asynchronicité en même temps, on obtient des performances plus hautes que la plupart des frameworks NodeJS et égales à celles du Go, qui est un langage compilé plus proche du C <a href="https://www.techempower.com/benchmarks/#section=data-r17&hw=ph&test=query&l=zijmkf-1" class="external-link" target="_blank">(tout ça grâce à Starlette)</a>.

      Crc32CSupplier[] functions = values();

      if (functions.length == 1) {
        // We're running under J2ObjC.
        return functions[0];
      }

      // We can't refer to JAVA_UTIL_ZIP directly at compile time because of J2ObjC.
      Crc32CSupplier javaUtilZip = functions[0];

      try {
        Class.forName("java.util.zip.CRC32C");
        return javaUtilZip;

-       unsigned char __uc;						      \
-     } *__u = __s;							      \
-     __uint8_t __c = (__uint8_t) (c);					      \
-									      \
-     /* This `switch' statement will be removed at compile-time.  */	      \
-     switch ((unsigned int) (n))					      \
-       {								      \
-       case 15:								      \
-	 __u->__ui = __c * 0x01010101;					      \
-	 __u = __extension__ ((void *) __u + 4);			      \

    protected long idleConnectionTimeout = 60 * 1000L; // 1min

    /** Pattern for matching HTTP redirect status codes */
    protected Pattern redirectHttpStatusPattern = Pattern.compile("[3][0-9][0-9]");

    /** Whether to use robots.txt disallow rules */
    protected boolean useRobotsTxtDisallows = true;

    /** Whether to use robots.txt allow rules */

			</plugin>
			<plugin>
				<groupId>com.github.hazendaz.maven</groupId>
				<artifactId>yuicompressor-maven-plugin</artifactId>
				<executions>
					<execution>
						<phase>compile</phase>
						<goals>
							<goal>compress</goal>
						</goals>
					</execution>
				</executions>
				<configuration>
					<nosuffix>true</nosuffix>

    }
  }

  private static <K extends Enum<K>, V> ImmutableMap<K, ? extends V> copyOfEnumMap(
      EnumMap<?, ? extends V> original) {
    @SuppressWarnings("unchecked") // the best we could do to make copyOf(Map) compile
    EnumMap<K, V> copy = new EnumMap<>((EnumMap<K, ? extends V>) original);
    for (Entry<K, V> entry : copy.entrySet()) {
      checkEntryNotNull(entry.getKey(), entry.getValue());
    }

	}
	return true
}

// symDefRef scans a line for potential text symbol definitions and
// references and writes symabis information to w.
//
// The symabis format is documented at
// cmd/compile/internal/ssagen.ReadSymABIs.
func (p *Parser) symDefRef(w io.Writer, word string, operands [][]lex.Token) {
	switch word {
	case "TEXT":
		// Defines text symbol in operands[0].
		if len(operands) > 0 {

	_ "unsafe" // for linkname
)

// Equal reports whether a and b
// are the same length and contain the same bytes.
// A nil argument is equivalent to an empty slice.
func Equal(a, b []byte) bool {
	// Neither cmd/compile nor gccgo allocates for these string conversions.
	return string(a) == string(b)
}

// Compare returns an integer comparing two byte slices lexicographically.
// The result will be 0 if a == b, -1 if a < b, and +1 if a > b.

   * decade that assumes that they can use checkNotNull for non-precondition checks. I had hoped to
   * take a principled stand on this, but the amount of such code is simply overwhelming. To avoid
   * creating a lot of compile errors that users would not find to be informative, we're giving in
   * and allowing callers to pass arguments that a nullness checker believes could be null.
   *