use fetch + keepalive

[gingerlime]

bumped into https://css-tricks.com/send-an-http-request-on-page-exit/ which describes a more reliable way to make sure requests are sent, even when the user leaves the page, using fetch + keepalive. Not sure if older browsers support it, or if we can use a shim. Another option is to replace the jQuery get with a fetch if it's available, or fallback on the plain JS get.

[gingerlime]

@joker-777 what do you think?

[joker-777]

@gingerlime Super interesting article. I wasn't aware of this behavior. In kenhub we use whatwg-fetch as a polyfill for fetch but we only load it when it doesn't exist. I would simply use fetch + keepalive and let the "developer" who uses Alephbet decide if he needs a polyfill or not.

[gingerlime]

So users can load a polyfill and the Alephbet code would automatically be able to use it?

[joker-777]

That's kind of what we do. We load this js code very early inline which checks if fetch or other methods exist and if one of them doesn't then we inject this script.

[joker-777]

Ah, and yes, once loaded, it is available for all js libraries.

[gingerlime]

Okay, but this feels like quite a lot of work for a user of Alephbet to have to worry about. I don't mind having a minimum-browser support, but then it needs to be documented clearly. Otherwise, some people might try to use it and will just get an error and won't necessarily know why. Or won't even know it can error in older browsers, and then their data is skewed.

The current JS fallback supports very old browsers, so we can also keep it like we do already (we check for jquery, but we can check if fetch is available instead). I think it's nicer to offer older-browser support out of the box if it's not super complicated for us to do that.

[joker-777]

I disagree. You would make the package bigger than necessary. Especially in the case of then we shouldn't worry about it because it is widely supported except for IE. If someone wants to support IE then they will use anyway polyfills which don't have to be by the way so "complicated" as we do it. They simply could include them and that's it.

[gingerlime]

We agree then :) I said it's fine to have a minimum browser support, but we just need to communicate it very clearly. And the code won't be much bigger even if we keep the fallback. We're talking about maybe 5 lines of code?

[joker-777]

Ah, ok, yes that's true. It doesn't save a lot of code.