cURL: Why do we even bother checking if we're online in this way?

[ilikenwf]

cURL: Why do we even bother checking if we're online in this way?

[xbmc-gh-bot]

Thank you for using Kodi and our issue tracker. This is your friendly Kodi GitHub bot 😉

It seems that you have not followed the template we provide and require for all bug reports (or have opened a roadmap item by accident). Please understand that following the template is mandatory and required for the team to be able handle the volume of open issues efficiently.

Please edit your issue message to follow our template and make sure to fill in all fields appropriately. The issue will be closed after one week has passed without satisfactory follow-up from your side.

This is an automatically generated message. If you believe it was sent in error, please say so and a team member will remove the "Ignored rules" label.

[ilikenwf]

The bot is annoying but I hate this function using a hard coded url.

[garbear]

The bot is annoying but I hate this function using a hard coded url.

I mean, "I hate this URL" is a pretty poor issue, better would be to create a PR with a more likeable URL. Discussion would ensue, we would weigh the costs/benefits, and would possibly align with a URL to your liking.

[ilikenwf]

Maybe I'll PR sometime, I get grumpy at night.

Even something like kodi.tv/online.txt or put an online.txt somewhere in a git repo instead...but meh.

[garbear]

kodi.tv/online.txt

Not gonna lie, I hate the M$ URL too. But putting it on our infra will for surely fail at some point and cause fallout which we don't need and can't handle. So we're relatively satisfied using the same system as basically the rest of the planet, checking a text file on MS's servers.

I'm increasingly interested in this issue as my company might use the microsoft text file in the interim while we switch between data ingestors (could be a few weeks). I really don't see a huge downside. Do you? Is there something we should be aware of, considering we fallback to w3c.org if the MS text file doesn't resolve?

[ilikenwf]

@garbear to be fair github is MS, but using github blends in more with most of our practices compared to other methods...

Firefox and other browsers tend to look for an http 204 response code, which doesn't return any body data but returns "204" in the header which is a "no response" code but still valid because of the 2XX value, making it super fast.

I'd rather use a mozilla connectivity test endpoint, or one hosted elsewhere..I just kinda freaked when I saw it because I run an all Linux/BSD network here and was perturbed to see an MS url in my firewall logs. Having it be a configurable, even if by advancedsettings.xml or something would be pretty nice.

That said, curl already has native handling for connection errors, so I don't know why we need to necessarily know, at least with remote connections, if we are "online" or not at all since curl will always gladly tell us anyway?

[ilikenwf]

Another option would be to instead to maybe use something like cloudflare's dns over https resolver to resolve kodi.tv or something...or just plain system DNS to resolve something.

While it doesn't necessarily provide that you're receiving data (an http 204 would) via the protocol it proves that you're connected to something.