Closed alshakero closed 6 years ago
Could you also try Chrome on Linux in Sauce? As that was the problematic combination
Passed too 😐
Not all of them.
Ha! So at least we got a trace, how to reproduce it locally, try in Chrome on Linux :)
Doing that now.
yay. passing on Sauce Chrome Linux. but not travis
What to do now haha?
Longer timeout fixed it for your local linux try? I suspect some nasty Chrome implementation bug like something on the edge of removing previous <link>
and de-duping.
Yes exactly. And I can't debug it! Because when I put debugger
before console.warn
it works without doing anything, and console.warn
always points to Sinon's spy when I can see it. Quantum physics we are talking.
@tomalec finally! Passing on Travis'es local Chrome and everywhere with 20ms delays.
Edit: I reorganized the files back now. They should pass 🙏
Ready for review.
Why is juicy-html
the only project that has a problem with that?
Because AFAIK it's the only one that depends on real network requests for tests, and on them being timely. The issue can be anywhere in polyserve
, the network stack, or even the OS. Because I have never seen a JS issue that is only reproducible in one OS version of Chrome.
I'll conduct one more expirement to rule out polyserve
.
Nailed it. Using http-server
for mocks solved the issue. https://github.com/Juicy/juicy-html/commits/use-http-serve-for-tests-mocks
However, I still lean to mocked ajax for speed and less-deps.
I gave it several shots with local Chrome and couldn't possibly come by the reason. Can we go for this?