GitLab

* commit '308df1d9':
  Remove DigiNotar Root CA

* commit '26085321':
  Remove DigiNotar Root CA

Change-Id: Ide84f6fa454eafb843e48cd21312875f387c2c13

* commit '623ec56a':
  Remove DigiNotar Root CA

* commit '203990d7':
  Remove DigiNotar Root CA

cd libcore/luni/src/main/files
git rm cacerts/c0cafbd2.0
./certimport.sh

Bug: 5232736
Change-Id: I2c78d90fe513ec6f4a0a76a3b4095d9be9345292

* commit '75aea4d5':
  Remove DigiNotar Root CA

cd libcore/luni/src/main/files
git rm cacerts/c0cafbd2.0
./certimport.sh

Bug: 5232736
Change-Id: I455a7c72a6d08cd0556fd4a64bb195e9e97dbfc0

Bug: 5232736
Change-Id: Ifb4c31d5ba994bc694aeabc218d5a6397a686769

Bug: http://b/5179775
Change-Id: Ie4b220eb61eb479b3d0bb894a3824fcdbc1a70b1

I don't believe the functionality this test was testing ever
worked. The new test makes that clear; the old test masked the
real behavior in a bunch of unnecessary complexity.

These tests demonstrate that query timeouts are not working!

Bug: http://b/5213614
Change-Id: If09e5d6204e06df7df4649953b69ddba0537fbf9

In classes that wrap another Socket, return the real FileDescriptor
from the wrapped Socket.

Bug: 5189186
Change-Id: I157feb6991def9110eaf0ea82365b6f5b95b9372

Change-Id: I243519eff36e31298b433069bf1661c57f9ef46a

Bug: 5205819
Change-Id: I06af9d59485126579266af062c3a81a8ac36b1f5

Change-Id: I18c826f70a820bdcd6bcb46a8af9038ba685fc30

It's a lot of work to cache partial responses properly. We've
punted on this for the built-in HttpResponseCache.

This test passed once, but only because we were using a stub
response cache that didn't interpret 206s properly.

Change-Id: I2ee5ae2d5dc9a9d52d499411b281d647e658f21d

I recently broke JarInputStream by changing Streams.readFully()
to unconditionally close the input stream when it was exhausted.
Although most streams won't return data after read() returns -1,
both JAR and ZIP streams can return data after a call to get the
next entry.

Change-Id: I2fcef44aaeae0ac27c9fa5a7a4996c5b1b794964

Change-Id: Ib1d256a1aa6cb74e09f183263bb0f3e40b103e4f

Change-Id: I7d8d683bd81fd96e826584cf502e418b29acab61

This was causing tests to fail under CTS.

Change-Id: Iaefa292321c77ca1644fc04655d6c9aa1cac9b88

This was causing the test to fail on CTS.

Change-Id: I43a55620e702d4693eb4f8df2791ec7bba8d8390

Change-Id: I27d13fe10153bfbb124f6259cf0b98ba2c442060

Bug: http://b/5171136
Change-Id: I58259863157df1301183d851d43bcfc6fcdb72c9

Our testing tools cannot normally test package-private APIs
because the tests are loaded in a different class loader than
the production classes. This is an unfortunate workaround.

Change-Id: Ic8d9d3088887bccceaf166959468cd52bfef9aae

Tests were failing in CTS because they were relying on the static
'st' member to be side effected by other tests.

This also makes some quick cleanups to style and names.

Change-Id: I93b9ce370c247c07f3521a8db3904ede86394ffe

The CTS test harness doesn't sanitize the preferences between tests
so we need to make sure it's clean in the test.

Change-Id: Idc80ac23f97431d31c902ad6a985ee8010a008f8

This weird, surprising behavior used the entire call stack as
context to lookup resources. This is the wrong solution to the
problem; context class loader is the right solution. And the
code is already using context class loader.

This change prevents some tests from failing with an NPE when
iterating the null class context.

Change-Id: Ic2419dc22f66a4fa8cceaee4746ef1a6d7d4680c