When the mean is 0, the coefficient of variation is calculated to be
NaN, which is not meaningful, so instead of printing "+-NaN%", just
don't print the CV at all.
Closes#908Closes#1444
Add two factory static functions to Injector: resolveAndCreate and
fromResolvedBindings.
We want to avoid resolution and flattening every time we create a new
injector. This commit allows the user to cache resolved bindings and
reuse them.
* Rename metrics, add `Time` suffix to all so that they are
more consistent
* iOS does not give us `gc` metrics, so they should not be reported
* Rename `scriptMicroAvg` into `microScriptTimeAvg`
* Rename previous `script` metric into `pureScriptTime` metric,
and keep `scriptTime` metric as the overall time, so that we still
have a shared metric across devices independent of the supported
browser features
* `microScriptTimeAvg` is now based on overall `scriptTime`, including
gc and render time.
* Move more shared DI tokens into `common_options` (previously
`sample_options`).
Closes#930
fixes#614
Asynchronous test should inject an AsyncTestCompleter:
Before:
it("async test", (done) => {
// ...
done();
});
After:
it("async test", inject([AsyncTestCompleter], (async) => {
// ...
async.done();
}));
Note: inject() is currently a function and the first parameter is the
array of DI tokens to inject as the test function parameters. This
construct is linked to Traceur limitations. The planned syntax is:
it("async test", @Inject (async: AsyncTestCompleter) => {
// ...
async.done();
});
Sam Rawlins
Yegor Jbanov
Tobias Bosch
Victor Berchet