Closed cvazac closed 6 years ago
@igrigorik It was certainly meant to link to something, but https://w3c.github.io/resource-timing/#cross-origin-resources didn't fit for me. The server timing bits tell me .serverTiming
shouldn't be exposed on ResourceTimingEntry objects. The link refers to durations being zeroed out. Or maybe the resource-timing bits should be updated and fix link.
Yeah, the behavior we ended up with for crossorigin server timing with no TAO seems significantly different than what we had for ResourceTiming.
Yeah, the behavior we ended up with for crossorigin server timing with no TAO seems significantly different than what we had for ResourceTiming.
I'm confused.. Per https://w3c.github.io/server-timing/#processing-model, we talk about doing a check against timing-allow-origin algorithm. So, it's same as RT — correct? That said, re-reading this particular update, I agree that cross-origin link is probably unnecessary.
As the spec reads today, a cross origin resource will have:
{
...
serverTiming: []
}
Previous versions of the spec had "zeroed out" entries:
{
...
serverTiming: [{
name: 'blah',
duration: 0,
description: ''
}, ... ]
}
Preview | Diff