Polling improvements: switch to faster host, accept error responses

[chrisbolin]

Faster host

We use ipv4.icanhazip.com currently because that's the standard for other JS polling libraries. But we can probably get much faster. Here are some quick tests:

--- ipv4.icanhazip.com ping statistics ---
round-trip min/avg/max/stddev = 154.028/159.056/188.548/11.154 ms
159 ms average

--- google.com ping statistics ---
round-trip min/avg/max/stddev = 23.071/24.103/26.079/0.830 ms
24 ms average

--- twitter.com ping statistics ---
round-trip min/avg/max/stddev = 51.824/52.745/53.685/0.731 ms
52 ms average

--- akamai.com ping statistics ---
round-trip min/avg/max/stddev = 39.591/40.428/41.625/0.620 ms
40 ms average

Can anyone think of a host to try that might beat google?

Accept error responses

#20 introduced a subtle change that we should look into more: looking at response status. We should accept (i.e. count as "online") any response, as long as it comes from the server. a 404 and a 500 both mean that we contacted the server, and are just as good as a 200.

[Francomgarcia]

Hey, it seems the library has an issue in windows. It shows this url http://ipv4.icanhazip.com/ can't be accessed.

[chrisbolin]

@Francomgarcia see #41

[chrisbolin]

You’re in luck! Check out the polling.url prop

…

On Fri, Jun 21, 2019 at 5:08 AM Krzysztof Węgliński < ***@***.***> wrote: nice, I'd extend it with customisable url so if I have i.e. api/status endpoint then I could use it — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#37>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAVQ7IBWVSTSREHKS76O6SDP3SZD5ANCNFSM4HCQRXIA> .