This is IA32 only for now.
Added a random value to each assembler instance (JIT cookie) to be used for constant splitting. Added safe versions of setting a register with an immediate value and for pushing an immediate value. Used these functions where user controlled immediate values could be emitted in the code stream. I also used it for immediates which are an argument number even though the number of formal arguments is currently limited to 16k.
I found no compares directly with user controlled constants.
I am not sure whether the test is that useful, but it might catch some changes missing constant splitting.
Review URL: http://codereview.chromium.org//7005031
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7868 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
The ScannerConstants class was originally static fields on the scanner class.
During creation of the stand-alone preparser and later isolates, it has been
moved into a separate class with a per-isolate instance.
It is used to hold caching unicode Predicate values.
This change renames the class to UnicodeCache, and passes a reference
to the instance down to methods that doesn't have an easy access to
an isolate (to avoid, e.g., having to do an Isolate::Current() for every
number parsed).
Review URL: http://codereview.chromium.org/6824071
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7584 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Patch by Dmitry Lomov.
pthreads implementations are free to reuse pthread_t (thread id) after
the thread has died. This change gets rid of ThreadHandle class and
replaces it with v8-managed thread identifiers.
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7575 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Correctly process failures which can be returned by Object::GetProperty
when performing GetRealNamedProperty* queries.
Callback properties can produce exceptions so we need to wrap access to them
into exception checks. However, despite of many other methods with exception
checks, property access doesn't mandatroy go via JavaScript and hence we
need to inject code to propagate exception to public API TryCatch handlers.
Review URL: http://codereview.chromium.org/6685087
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7548 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
We can only call malloc/free once per group and we can avoid scanning
through a list of NULLs if we keep unprocessed groups in the beginning.
I also changed the internal representation of implicit references to
hold a handle to the parent (instead of a direct pointer). The
prologue callback must not trigger a GC, but it's better to be safe.
Review URL: http://codereview.chromium.org/6800003
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7521 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
While trying to fix Mac and Windows versions for this change:
http://codereview.chromium.org/6771047/, I figured out, that we
already store an isolate in StackFrameIterator, so we can use it in
frame objects, instead of requiring it from caller.
I've changed iterators usage to the following scheme: whenever a
caller maintains an isolate pointer, it just passes it to stack
iterator, and no more worries about passing it to frame content
accessors. If a caller uses current isolate, it can omit passing it
to iterator, in this case, an iterator will use the current isolate,
too.
There was a special case with LiveEdit, which creates
detached copies of frame objects.
R=vitalyr@chromium.org
BUG=none
TEST=none
Review URL: http://codereview.chromium.org/6794019
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7499 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
It should now be possible to build the preparser using 'scons preparser' in both release and debug modes.
Remove v8.h include from scanner-base.h and other files.
Remove NativeAllocationChecker and all of its kind.
Moved Isolate::PreallocatedStorage* to isolate.cc
Review URL: http://codereview.chromium.org/6749029
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7413 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This commit adds current working versions of assembler, macro-assembler,
disassembler, and simulator.
All other mips arch files are replaced with stubbed-out versions that
will build.
Arch independent files are updated as needed to support building and
running mips.
The only test is cctest/test-assembler-mips, and this passes on the
simulator and on mips hardware.
TEST=none
BUG=none
Patch by Paul Lind from MIPS.
Review URL: http://codereview.chromium.org/6730029/
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7388 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This patch adds common infrastructure for fast TLS support and
implementation on win32. More implementations will be added soon.
Fast TLS is controlled by V8_FAST_TLS define which is enabled by
default in our gyp and scons builds. The scons build has
fasttls={on,off} option so that we can see the effects of slow TLS
when needed.
Review URL: http://codereview.chromium.org/6696112
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7375 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Access to an ExternalReference in non-serializable code will try to use
an offset relative to the root-array register.
Since the root-array is in the Heap object, and the Heap object is in
the Isolate object, there's a good chance that any external data field
is within a 32-bit offset of the root array register.
It falls back on the original behavior if the serializer is enabled,
if the root register isn't initialized or if the offset is not representable
as a 32-bit value.
Review URL: http://codereview.chromium.org/6716018
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7315 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Because we run all tests three times with different variant flags (to
test crankshaft) we might end up in a situation where we try to write
to the same serilization file from two different threads
simultaneously. The patch concats the variant flags at the end of the
serialization file name.
Review URL: http://codereview.chromium.org/6688068
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7285 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Correctly process failures which can be returned by Object::GetProperty
when performing GetRealNamedProperty* queries.
Callback properties can produce exceptions so we need to wrap access to them
into exception checks. However, despite of many other methods with exception
checks, property access doesn't mandatroy go via JavaScript and hence we
need to inject code to propagate exception to public API TryCatch handlers.
Review URL: http://codereview.chromium.org/6397011
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7258 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
we know that both sides are Smi and those where we don't. Fix inlined
symbol table probes to cope with strings, undefined and null (indicating
a deleted entry). Some changes to other architectures that were found
with the new asserts.
Review URL: http://codereview.chromium.org/6682026
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7172 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Worth mentioning:
- Specialized versions of pixel array and store/loads inside the generic stubs have been removed, since to have parity for all external arrays, 8 different versions would have to be inlined/checked.
- There's a new constant in v8.h for external arrays with pixel array elements.
Review URL: http://codereview.chromium.org/6546036
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@7106 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
The main issue was due to multiple recompilations of functions. Now
code objects are grouped by function using SFI object address.
JSFunction objects are no longer tracked, instead we track SFI object
moves. To pick a correct code version, we now sample return addresses
instead of JSFunction addresses.
tools/{linux|mac|windows}-tickprocessor scripts differentiate
between code optimization states for the same function
(using * and ~ prefixes introduced earlier).
DevTools CPU profiler treats all variants of function code as
a single function.
ll_prof treats each optimized variant as a separate entry, because
it can disassemble each one of them.
tickprocessor.py not updated -- it is deprecated and will be removed.
BUG=v8/1087,b/3178160
TEST=all existing tests pass, including Chromium layout tests
Review URL: http://codereview.chromium.org/6551011
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6902 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Object.preventExtensions can currently be used cross-domain. With this
change we follow firefox (IE9 has our current behaviour). In addition
this includes a regression test for 1027 and access tests for
Object.seal and Object.freeze.
Review URL: http://codereview.chromium.org/6534019
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6848 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Call accessors in the global object prototype when initializing global
variables. Function declarations are special cased for compatibility
with Safari and setters are not called for them. If this special
casing was not done webkit layout tests would fail.
Make the declaration of global const variables in the presence of
callbacks a redeclaration error.
Handle const context slot declarations conflicting with a CALLBACK as
a redeclaration error. That is, unless it is on a context extension
object which is not a real object and therefore conceptually have no
accessors in prototype chains. Accessors in prototype chains of
context extension objects are explicitly ignored in SetProperty.
Review URL: http://codereview.chromium.org/6534029
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6846 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
In some circumstances, try/finally block can actually catch the exception:
function f() {
try {
throw 42;
} finally {
return 0;
}
}
Therefore when propagating exception to v8::TryCatch, we must be sure
there is no try/finally blocks as well.
When bulding the messages we should be more conservative and expect that
any v8::TryCatch with no JS try/catch in between can potentionally
be the right exception handler.
Plus various minor refactorings.
BUG=1147
TEST=cctest/test-api/TryCatchAndFinallyHidingException, cctest/test-api/TryCatchAndFinally
Review URL: http://codereview.chromium.org/6526016
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6809 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Strict mode assignment to undefined reference.
Simple assignments (x = <value>) use CODE_TARGET_CONTEXT.
StoreIC stores its own strictness in extra_ic_state.
The strcitness is propagated as further ic stubs are generated.
Details:
* ReferenceError on assignment to non-resolvable reference in strict mode.
* Fix es5conform test expectation file.
* Add es5conform test suite into .gitignore.
* Fix Xcode project.
* Change implemented in virtual frame code generator, as well as full-codegen
for all architectures.
* Fix debugger test.
* Fix comment for CODE_TARGET_CONTEXT
* Implement remaining StoreIC stubs to be strict mode aware.
* Trace extra_ic_state() for ic code stubs.
Code Review URL: http://codereview.chromium.org/6474026/
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6760 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Strict mode assignment to undefined reference.
Simple assignments (x = <value>) use CODE_TARGET_CONTEXT.
StoreIC stores its own strictness in extra_ic_state.
The strcitness is propagated as further ic stubs are generated.
Details:
* ReferenceError on assignment to non-resolvable reference in strict mode.
* Fix es5conform test expectation file.
* Add es5conform test suite into .gitignore.
* Fix Xcode project.
* Change implemented in virtual frame code generator, as well as full-codegen
for all architectures.
* Fix debugger test.
* Fix comment for CODE_TARGET_CONTEXT
* Implement remaining StoreIC stubs to be strict mode aware.
* Trace extra_ic_state() for ic code stubs.
Code Review URL: http://codereview.chromium.org/6474026/
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6756 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
It made value of allowed_access_type[v8::ACCESS_KEYS] be a wild value. On most of
platforms it was 0 and tests passed. But on ARM (and on ia32 if you alter test a bit)
it could become true and hence allow enumeration of properties.
Review URL: http://codereview.chromium.org/6334089
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6618 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Current approach returns undefined descriptor if caller is not granted v8::HAS_ACCESS.
If the caller has v8::HAS_ACCESS, for no JS accessors regular v8::GET_ACCESS check is
performed and value property of the descriptor is set to undefined if caller doesn't
have proper access. For JS accessors both v8::GET_ACCESS and v8::SET_ACCESS are checked
and affect if getter and setter would be stored in the descriptor.
Review URL: http://codereview.chromium.org/6286020
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6592 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Record a safepoint with a deoptimization id for throw in optimized code. We
don't seem to much care what the AST ID is because we will not be using it
for lazy deoptimization (throw doesn't return to the point of throw). For
hygiene we use the actual ID of the throw expression. Throw is no longer a
control-flow instruction, but it's followed by an unconditional abnormal
exit. This is required to insert a simulate between the throw and the exit.
Make our optimized treatment of Function.prototype.apply act like a call and
have side effects. This ensures that it will get a lazy deoptimization
environment. Use that deoptimization ID in the safepoint for the call.
Deleting a property was also missing a deoptimization ID, though there was a
deoptimization environment assigned to the instruction. Record the
environment and use the deoptimization ID at the safepoint.
Review URL: http://codereview.chromium.org/6250105
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6576 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
With crankshaft, a code object can change its optimizability: it can start as
optimizable code object, but later we can find out it was a bad idea to
optimize it. Alas, currently we don't have a proper event to communicate
this back to logger. Hence we temporary allow a code object to be viewed
as optimizable judging from logs while being unoptimizable judging from
heap traversal.
Review URL: http://codereview.chromium.org/6250054
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6553 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
other situations.
Do not use overwritten Object.prototype.hasOwnProperty and
Array.prototype.pop. Do not use split and join in the error formatting
implementation. They are too big to control and their generality is
not needed.
Review URL: http://codereview.chromium.org/6287041
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6552 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Fix a bug in the --debug-code alignment check in the C entry stub.
Don't force the --debug-code flag in the ARM disassembler tests. The framework does support passing flags and the test runner will when running tests in debug mode.
Skip some deserialization tests which crashes from time to time.
Review URL: http://codereview.chromium.org/6393007
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6484 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
ReferenceError.prototype.__proto__ which will make "error instanceof
Error" fail. However, the ReferenceError.prototype object itself
cannot be modified. Therefore, the error checks must check for
concrete error instances instead of only checking for Error.
Review URL: http://codereview.chromium.org/6388003
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6450 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
conversions for external array types, which implement the Typed Array
spec. (Revision of http://codereview.chromium.org/6315004 .)
Prefer SSE2 code path on x86 processors. Non-SSE2 processors now make
a slow runtime call for float-to-int conversions. Use SSE3 for 32-bit
signed and unsigned int array types where possible.
The movement of code from ic-arm.cc to stub-cache-arm.cc caused the
VFP3 code path to be tested for the first time. Fixed bugs in the
register usage and in the constant value stored into integer arrays
for NaN and +/-Infinity.
Added new truncation test to test-api.cc. Storage of NaN and +/-Inf
was already covered. Ran unit tests on x86, x64 and ARM simulator.
Tested ia32 and x64 code in Chromium on Mac and Linux respectively
with Typed Array unit tests and WebGL content.
BUG=http://code.google.com/p/chromium/issues/detail?id=50972
TEST=test-api/ExternalArrays
Review URL: http://codereview.chromium.org/6303012
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6431 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This change includes support for safepointtables and adding deoptimization info (but not for deoptimizing).
Implemented crankshaft functions:
CallCode
GenerateSafepointTable
RegisterEnvironmentForDeoptimization
EmitGoto
This change allows us to compile very simple functions with crankshaft:
An empty function
A function returning a constant.
A function returning a parameter.
There is 6 disabled tests that require us to be able to deoptimize
which is currently not supported.
Review URL: http://codereview.chromium.org/6310009
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6350 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
On 32-bit platforms any pointer with 0 as LSB can be wrapped into Smi. However, on 64-bit
platforms it's currently not the case as x64 Smis must have 0s in lower 32 bit word.
Even worse, macroassembler Move instruction will try to fetch integer value from Smi
and will shift by 32 bits to the right rendering stored pointer incorrect.
BUG=v8:1037
Review URL: http://codereview.chromium.org/6119009
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6301 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This patch enables two new flags for the tools/test.py script;
--shard-count - giving the ability to split the tests to be run
into shard-count chunks.
--shard-run - giving the ability to specify which of the shards to actually run.
Example
tools/test.py -j15 --shard-count=2 --shard-run=1 mozilla
would split the mozilla tests into two chunks and run the tests in the first chunk
Running:
tools/test.py -j15 --shard-count=2 --shard-run=1 mozilla
tools/test.py -j15 --shard-count=2 --shard-run=2 mozilla
is equivalent (in terms of test coverage) of just running:
tools/test.py -j15 mozilla
In addition, tests are now sorted before they are returned from the
test specific ListTests methods (sputnik and mozilla tests where
already sorted before they where returned).
This change is needed to split a single test suite over two slaves on
the waterfall.
Review URL: http://codereview.chromium.org/6127003
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6248 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
1. Added gdb style debugger commands (and their shortcuts) for d8.
These include:
- s[tep] : step into the current statement.
- s[tep]i[n]: step into the current statement with the minimum step.
- n[ext] : step to the next statement.
- fin[ish] : step out of the current function.
- cond : setting conditions on breakpoints.
- d[elete] : deletes breakpoints.
- en[able]|dis[able]: enables/disables breakpoints including
exception breakpoints.
- ignore : ignores a breakpoint for a specified period.
- inf[o] ar[gs] : info on arguments of the current function.
- inf[o] lo[cals] : info on local vars of the current function.
- inf[o] br[eakpoints] : info on breakpoints.
- l[ist] : similar to source, but allows the user to continually
dump subsequent lines of source code either in the
forward or backward direction.
- quit / exit / disconnect : terminates the remote debugger
session.
NOTE: Active breakpoints will automatically be disabled when
the remote debugger detaches. This allows v8 to continue to
run without worrying about a loss of a debugger session.
2. Added support for breaking the debugger by simply typing ENTER.
The break command is now optional.
3. Once the debugger is broken, the user can now just type ENTER
to repeat the last command. This is useful to functionality that
needs to be invoked repeatedly e.g. step, list.
4. Added more verbose descriptions in d8's help.
5. Fixed a line and column number offset bug in the listing of breakpoint
line and column numbers.
6. Added a gc command to allow GCs to be requested from the debugger
interface. The plumbing for requesting different types of GCs is
there, but the underlying implementation currently only triggers a
full mark-compact GC. The command also returns the before and after
sizes of the heap.
7. Added trace json, and flags commands that are not published in help.
trace json is used for tracing the debugger packets send from and
received by d8. flags is for setting v8 flags. These are useful for
people debugging v8 itself, but not necessarily users of v8.
8. Added the ability to enable and disable break on all / uncaught
exceptions in to d8.
9. Added a fix to prevent the Debugger Agent from being re-instantiated
if one already exists.
10. Added the ability to filter results of the script command by matching
text or numbers on the results.
11. Added v8 flags to enable/disable the sending of debugger BeforeCompile,
AfterCompile, and ScriptCollected events.
12. Fixed some undefined value bugs that resulted in v8 or the debugger
failing.
13. Added a few minor WEBOS__ customizations (analogous to ANDROID
customizations).
Patch by Mark Lam from Hewlett-Packard Development Company, LP
Review URL: http://codereview.chromium.org/5980006
Review URL: http://codereview.chromium.org/6086010
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6200 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
This is based on the structore used in chromium with a script wrapping the call to gyp itself and the default processing of common.gypi.
It is possible to build all our targets on Intel Linux for all architectures (ia32, x64 and ARM simulator). When this is committed I wil take a look at Windows.
See the README.txt file in the changelist for the current way of using it.
Review URL: http://codereview.chromium.org/5701001
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@6000 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
As taking a snapshot of a large heap takes noticeable time, it's
good to be able to monitor and control it.
The change itself is small, big code deletes and additions are in
fact moves. The only significant change is simplification of
approximated retained sizes calculation algorithm.
Review URL: http://codereview.chromium.org/5687003
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5978 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Approximation is done by building a dominators tree for the heap graph.
Dominator nodes and retained sizes are serialized into JSON.
Removed:
- reachable size (it is useless, after all);
- HeapEntryCalculatedData (size is now stored in the node, retaining
paths in a hash map);
Review URL: http://codereview.chromium.org/5154007
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5867 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Otherwise, retaned memory sizes are not precise. This increases size
of heap snapshot, I will deal with this later. Heap objects and
references previously missing in snapshot are now marked as 'hidden'.
That means, they not shown to user, but participate in sizes
calculation.
Other small changes:
- added 'shortcut' graph edges: e.g. to pin global objects on top
level;
- meta-information in JSON snapshot is no more double encoded.
Review URL: http://codereview.chromium.org/5139002
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5849 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Make checks.h not depend on flags.h or global.h (or anything else except
include/v8stdint.h). Only checks.cc has the dependencies (so another
implementation of checks.cc can be provided by the preparser).
Now files depending on checks.h (using ASSERT macros) can include it
directly without depending on all of v8.
Review URL: http://codereview.chromium.org/4576001
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5775 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Instead of relying on itimer signals from kernel, send them
ourselves from a separate thread. This disables an ability
to profile multiple VM threads on Linux, but it anyway doesn't
work on other platforms, so we need a common solution for
it (issue 913 created to track this).
Review URL: http://codereview.chromium.org/4000007
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5711 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
Don't use floating-point operations on Linux,x86 to compute strtod. Since the
floating-point stack on Linux is set to 80bit double rounding may occure.
When falling back to gay_strtod append several '0's so that Gay doesn't take
the same shortcut either.
BUG=
TEST=
Review URL: http://codereview.chromium.org/3851003
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5650 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
HeapNumbers do consume memory, so it's worth dumping them. However, we
don't dump their values, as they are not as self-descriptive as values
of strings, and they will increase snapshot size. Storing heap numbers
values can be added if we will feel a sufficient demand for that.
InternalFields are used, e.g. for storing references to DOM nodes
event handlers.
Review URL: http://codereview.chromium.org/3769007
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5635 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
