GitLab

Aside from being an unjustified optimization, users have reported
problems with this in the wild. This cache has already been removed
in master.

* changes:
  Cache NumberFormat and DecimalFormatSymbols objects in a ThreadLocal, so they can be reused between multiple instances of Formatter on the same thread.  This speeds up my unscientific benchmark (a number of printouts involved in a debugging diagnostics output) by 3x, and should have a similar impact on anyone who uses String.format(), PrintWriter.format(), and the like.

* changes:
  Fix Formatter.format's handling of Formattable.

* changes:
  Consistently use Formatter's cached NumberFormat.

so they can be reused between multiple instances of Formatter on the
same thread.  This speeds up my unscientific benchmark (a number of
printouts involved in a debugging diagnostics output) by 3x, and
should have a similar impact on anyone who uses String.format(),
PrintWriter.format(), and the like.

This doesn't make much difference to performance, but it is
slightly faster and I think it reads better too. Proof of the
latter is the fact that the rewritten form accidentally fixed
bug 1767: the old code was storing up literal text until after
handling the next format specifier, which is wrong if the
format specifier has side effects caused by the use of
Formattable.

(I don't plan on doing any more on the performance bug for now,
though I note that %g allocates and manipulates BigDecimal
instances, which would be worth looking at when we get round to
the bug that causes it to fail a harmony test.)

Bug: 1476, 1767

%f is a lot more expensive than it should be because we're not using the cached
NumberFormat (as we are for %d). Running the microbenchmark I added to the
bug (times in ms, on a Cortex A8):

                                old       new
      new Formatter %f         1732       811
      String.format %d          635       651
      String.format %f         1752       900
      reuse Formatter %f       1521       188
      Double.toString +         149       148
      Double.toString append     33        33
      StringBuilder.append      143       139

Bug: 1476

This cuts out some unnecessarily verbose dalvikvm chatter, notably:

  Trying to load lib /system/lib/librs_jni.so 0x0
  Added shared lib /system/lib/librs_jni.so 0x0

These messages can be useful for people trying to get their apps to
work with the NDK, so I'm only suppressing them when the path starts
with "/system".  The result is that you can boot the system and run all
standard apps without seeing them, but we'll still see app-private libs
being loaded.

Also LOGI->LOGV for "Debugger thread not active, ignoring DDM send",
which seemed to be firing on startup for APp NaMe events.  Ditto for
"Splitting out new zygote heap", which only happens once, but doesn't
strike me as a particularly useful thing to log.

Merge commit '3dc6a81182178a8ac9570a7679cf24b6a2665a0b' into eclair-mr2

* commit '3dc6a81182178a8ac9570a7679cf24b6a2665a0b':
  DO NOT MERGE. CTS test cleanup. Mark tests as KnownFailures and BrokenTests.

See bug 2186927

Merge commit '3238cc2a901ed6e0e3bb11fa601913c2ed09ddb7' into eclair-mr2

* commit '3238cc2a901ed6e0e3bb11fa601913c2ed09ddb7':
  Remove tests.api.java.io.FileTest#test_delete as known failure.

Merge commit '98f0612b22111dd9e2d23b171b9059582cc44991' into eclair-mr2

* commit '98f0612b22111dd9e2d23b171b9059582cc44991':
  Fix ClassLoaderTest to work on DEXPREOPT builds.

Merge commit '32f8982116e5155bc3c75c86e3c861cab1a848be' into eclair

* commit '32f8982116e5155bc3c75c86e3c861cab1a848be':
  Remove tests.api.java.io.FileTest#test_delete as known failure.

Merge commit 'c4d33d9d89444d86808be7072e0548007d98a49f' into eclair

* commit 'c4d33d9d89444d86808be7072e0548007d98a49f':
  Fix ClassLoaderTest to work on DEXPREOPT builds.

* changes:
  Make the socket test accept ::1 as a valid result for InetAddress.getByName(null) if the system is set to prefer IPv6 addresses.

InetAddress.getByName(null) if the system is set to prefer IPv6 addresses.

Change-Id: I96cd6f5b632994b73af13e341cf6f684f6eee4cd

1. Make native code properly set network interface indexes that are used for
   new-style multicast socket options.
2. For IP_MULTICAST_IF, check that the address is IPv4, not the socket.
3. Remove a @KnownFailure annotation for a test that now passes.

This removes one test error and changes it into a more benign failure. The
remaining test failures seem to be due to Linux bening permissive about which
interfaces you can receive multicast traffic on, which is probably good enough
for now.

Change-Id: Id1fddee338addcfce821672b983485c7dd4983d4

* changes:
  Fixes for socket options on multicast sockets.

1. Properly pass get/setsockopt the pointer to the socket option instead of the
   pointer to the pointer to the option. This was not caught at compile time
   because it's a void *.
2. Handle IPv4 multicast addresses on IPv6 sockets. This is important because
   current devices create IPv6 sockets by default.
3. Use the proper options for IPv6 multicast (i.e., IPV6_{ADD,DROP}_MEMBERSHIP
   instead of IP_{ADD,DROP}_MEMBERSHIP)
4. Use integers instead of bytes when getting or setting the multicast TTL
   because that's what the Linux kernel uses.

These fix 10 of the 11 MulticastSocketTest failures. Also, minor changes:

1. Add ifdefd-out logging functions for get/setsockopt.
2. Change all instances of IPPROTO_{IP,IPv6} to SOL_{IP,IPV6} in get/setsockopt
   calls. Even though the values are the same (so the code worked), this way is
   more correct.

Change-Id: Iea75a523d7e71f0b361a42c0e39d3ef075dc7ff4

* changes:
  Remove @KnownFailure tags for tests that pass.

Also cleanup tests.api.java.io.SerializationStressTest4#test_writeObject_Proxy

* changes:
  Fixing a lame test that was relying on hashmap iteration order.

* changes:
  Fixing a Harmony regression in ObjectInputStream.

We check the field type against the instance being
populated and not the field descriptor on the wire.
The root cause is a bug in Harmony which we should
send upstream.

* changes:
  Fixing tests that exercise SecurityManager plus stack inspection.

The tests were both broken previously. They both relied on having
specific bootloader classes in their call stack.

* changes:
  Fix BigInteger math bugs.

This initializes the internal representation before doing left shifts.
I'd originally missed this in the first Harmony update; change 20002.

* changes:
  DO NOT MERGE: Cleaning up PipedInputStream

See bug 2021674 for discussion. I searched Google Code Search to
sample which of these APIs are being used. None of these APIs
were used by applications, so I'm quite confident that removing
them will not break applications or upset developers.

These APIs were neither formerly documented, nor intended for use
by application developers. I am deprecating them now to discourage
their use going forward. We should remove them in a future release.

Functions that could call dvmAbort() from more than one place were
merging the calls together, which is great until you need to decode a
native stack trace.

For 2144313.

* changes:
  Use native code to convert strings to IP addresses.

1. Add an ipStringToByteArray method to the OSNetworkSystem interface and
   provide a native implementation for it.
2. Change InetAddress to use the new method.
3. Remove  a mid-sized chunk of Inet6Util, which is now redundant.
4. Remove the KnownFailure annotations from the InetAddress, InetAddress and
   Inet6Address tests that expected 1.2.3, 1.3 and 3 to be valid IPv4
   addresses. These tests now pass again.
5. Removed an unused native method.
6. Added a comment to the top of the native socket implementation to the effect
   that the file is significantly different from harmony.

Change-Id: I3665f82b00ebc089e9133cc6166dda5a99fa10e4

Some content changes inspired by external developer feedback, plus some
corrections of long-standing typos.

* changes:
  Another doc update.

* changes:
  Fixed LinkedHashMap bug 2121546 Also made minor improvements in LinkedHashMap and NegativeCache. (The "opportunities for improvement" were discovered while investigating the bug.)