jgraham, could you hold off on merging pending a reply to web-platform-tests/wpt #21762#issuecomment-585661066 ?

Ms2ger: Sure

Although I didn't think that WebKit was actually syncing this stuff

Define 'this stuff'? They manually sync a sub-set of WPT and run it... somehow (I'm actually not clear on that part), so yes big changes like these can break them. clopez has a lot of experience with their WPT sync; clopez do you know what work (if any?) WebKit would need to do to support the new assert_throws_js/assert_throws_dom/etc APIs that were added to testharness.js and which we've now converted many tests to use?

Presumably webkit would need to sync all the tests they import (that use a_t/p_r) and then they could pull in the change to remove it from th.js

I assume that is the case, but think it is sensible to clarify (glad you thought about it!)

Right, but the failure case is if they do something weird like sync testharness.js but don't sync tests

Which seems like a bad idea in general

jgraham: which tbf is what Blink does ;P

Wait what? We sync both together...

The weird part is that we also have an internal copy of testharness.js, which is used for non-exported tests

'weird'

gsnedders: I thought blink did the opposite

Didn't update infra when doing syncs

Depends what piece of infra you mean. #itscomplicated

Once we get wptrunner in Chromium, our WPT tests should mostly nearly-always use sync'd infra...

how can I change the default ports used for the various servers during the WPT tests?

jgraham: ^

jdm: github.com/web-platform-tests/wpt/b…rom-local-system.md#via-the-browser

config.json file

thanks!