Jacob Page

Here's a weird one that took 630ms. The resource type is `TCPCONNECTWRAP`: ``` at GetAddrInfoReqWrap.emitLookup [as callback] (net.js:1019:9), at GetAddrInfoReqWrap.onlookup [as oncomplete] (dns.js:61:10) ``` Maybe DNS is getting back an...

Any updates on this? I'm looking to upgrade some existing tests that use `supertest` so they use http2.

I've tested out these changes on my HTTP2 server and @michael42's suggestions seem to be working. PR: https://github.com/expressjs/compression/pull/155. Can you take a look, @dougwilson?

> The main issues run into is that the proposed fix _seems_ fine on the surface, but only if you are trying to just make it not hard crash. There...

@maritz, @dougwilson: need any help with this PR? I'd definitely prefer using this over rolling our own compression support for HTTP2.

@maritz: do you have any hints on what your app does that might be provoking those failures? So far, I've been unable to reproduce that failure. It may be that...

I'll be doing a bunch of testing of your published fork within our app soon; maybe that will be sufficient validation. Will report back.

+1 for this. It'd be very helpful for enforcing organizational coding styles.

In scrutinizing further, it seems clear that it's in the defining of indexer properties where this problem is occurring: https://github.com/Financial-Times/polyfill-service/blob/1f45e5b95625180117d5b2f1979d8e217bc791a4/packages/polyfill-library/polyfills/_DOMTokenList/polyfill.js#L35-L41

`replaceChild` isn't working either for replacing the text node; it's still serialized with the original value. Neither does changing `textContent`.