It's a third party web application. It seems tuned for Safari but most people play it using desktop or mobile Chrome I maintain a desktop extension that fixes a bunch of stuff.
-
-
Replying to @antumbral
Got it. So the code you're loading is from the extension?
2 replies 0 retweets 0 likes -
Replying to @slightlylate
No, this is all their code. I'm running some logic to inject the prefetch/preload elements. For some reason their XHRs aren't picking up the prefetched data from the memory cache.
1 reply 0 retweets 1 like -
Replying to @antumbral @slightlylate
I'm injecting the prefetch elements from the page's execution context (not a content script) so I don't think sandboxing should be in play. I checked the headers in devtools and they all match (with the exception that their xhrs have an 'origin' header)
1 reply 0 retweets 0 likes -
Replying to @antumbral @slightlylate
I do hook http://xhr.open and xhr.send so if there's some way I can modify their XHR invocations to make preload actually work, that'd be great. I didn't see anything in docs anywhere.
1 reply 0 retweets 0 likes -
Replying to @antumbral @slightlylate
I previously tried caching and prefetching everything myself using idb, but redirecting their requests to blob: and data: urls was slower (blobs hit disk cache, data urls are slow to decode)
1 reply 0 retweets 0 likes -
Replying to @antumbral @slightlylate
Did you come up with any ideas? Should I file a bug about prefetch not working?
1 reply 0 retweets 0 likes -
Replying to @antumbral
I suspect you're past the preload scanner by the time you inject that, so perhaps a bug but hard to tell. Big issue here is cache speed.
1 reply 0 retweets 0 likes -
Replying to @slightlylate
Odd since documentation out there insists that doing preload this way should work. I inject them very, very early - hundreds of msec before they actually xhr to get the content.
1 reply 0 retweets 0 likes -
Replying to @antumbral @slightlylate
It preloads all of my injected preloads all at once, too - the whole set completes in like 50ms. So it's doing the preload very efficiently, but then it throws it out or doesn't use it.
1 reply 0 retweets 1 like
Worth filing a bug. I'll try to raise cache perf issue again with that team. There has been some turnover/priority changes there so will need to figure out who owns.
-
-
Replying to @slightlylate
OK. I'll have to see if I can come up with a reliable repro for this that isn't complicated.
1 reply 0 retweets 1 like -
Replying to @antumbral @slightlylate
At this point I am convinced link rel=preload does not work for fetch and I'm not sure it ever worked. Do you know how comprehensive the test coverage is for this?
1 reply 0 retweets 0 likes - 5 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
& Web Standards TL; Blink API OWNER
Named PWAs w/
DMs open. Tweets my own; press@google.com for official comms.