Previously, DeclarationResolver matched declarations to elements using their offsets in the file. This meant that it could not work with element models built from a summary, since those element models are not guaranteed to contain offsets. Now, DeclarationResolver assumes that elements appear in the same order in the element model as they do in the file, so it can just match them up using their order.
DeclarationResolver continues to verify element names as it does its matching, so if the ordering of elements ever fails to match up with the ordering of declarations, an exception is almost certain to be thrown. (The only time it wouldn't be is if the mismatch can't be detected due to multiple elements having the same name).
The new implementation is ~1.6x faster than the previous one.
R=brianwilkerson@google.com, scheglov@google.com
Review URL: https://codereview.chromium.org/2419633002 .
The catch entry block has all locals in fixed locations
(Rj) where j = kNumberOfRegisters - i for parameter i.
This means we reserve a range of DBC registers at the top-end of the frame.
Those registers are blocked for general allocation to avoid any overlap
with the rest of the registers that are allocated from the bottom.
Each optimized frame with a try-catch will be kNumberOfRegisters wide.
BUG=
R=vegorov@google.com
Review URL: https://codereview.chromium.org/2388093003 .
My initial goal was just to move some of the HTML tests off unittest,
but that ended up slurping in some other changes as well. I haven't
finished going through all of the tests, but this puts the
infrastructure in place and fixes a batch of them. I'll do more next.
Changes:
- Add a "minitest" package. This shims just enough of the unittest/test
API to make it easy to swap it out for unittest without having to
rework the entire test.
- Go through the HTML tests from A-J and change them to use minitest
when possible. The ones that aren't possible use asynchrony.
Minitest doesn't support async tests (yet, maybe).
- Triage those tests and enable ones that are passing. Mark them failing
if failing. Filed a couple of tracking bugs for the common issues,
but for other cases, I just left a comment.
- In the process of this, I went ahead and make some of the affected
tests strong mode clean if they weren't already and it was easy.
- Synchronize changes between dev_compiler's fork of expect and
pkg/expect. We'll want to unfork those at some point. D-:
This gets dev_compiler from 1985 tests to 2042.
R=jacobr@google.com, vsm@google.com
Review URL: https://codereview.chromium.org/2413073002 .
- [x] Make `reloadSources` RPC public.
- [x] Add an optional 'pause' parameter to `reloadSources`. When set to true, the isolate will enter the debug pause loop immediately after reloading sources and before resuming execution. This makes it possible for debugger clients to set breakpoints before resuming.
- [x] Add reload_sources_test which tests pausing after reload.
- [x] Fix a bug in `GetVarDescriptors` where a null code object could be used.
R=turnidge@google.com
Review URL: https://codereview.chromium.org/2411153002 .
This makes the following improvements:
* Always show the call that was made, with arguments.
* ...and show the actual arguments in more cases.
* Be less confusing in the case of a call on the null receiver.
* Be less redundant in the output.
* Report when a constructor is called with the wrong number of arguments.
R=rmacnak@google.com
Review URL: https://codereview.chromium.org/2405353002 .
Wait for concurrent sweep tasks before verifying points, as building the allocation set expects unmarked objects.
Make assertion okay with a missing stack map at entry, which transiently exists for OSR, as the frame looks the same as unoptimized code here.
Fixes#13608Fixes#23683
Reproduces issue #26927R=fschneider@google.com
Review URL: https://codereview.chromium.org/2411453003 .
Flutter layout benchmark 14k us -> 10k us. ~30%
improvement in checked mode. There still appear
to be a lot of checks that should be hitting the
cache, so there may be other fast-paths needed
as well.
R=fschneider@google.com
Review URL: https://codereview.chromium.org/2404163004 .
If there is no override of noSuchMethod in any class, we can
propagate the receiver type downwards from calls that call a unique
selector.
This speeds up dart2js by around 3%, a particle simulation benchmark by around 10%.
BUG=
R=rmacnak@google.com
Review URL: https://codereview.chromium.org/2412653002 .
