Support for https endpoints in witnesses and controllers

[rodolfomiranda]

No description provided.

[AlexAndrei98]

tests are passing!

[jasoncolburne]

https://github.com/WebOfTrust/keripy/pull/555/files#diff-a4229d5895c922898cdeb03b5bb5c550ea2bf933cef6772027ffa763936ca630R828-R835 looks like it may need mods

[jasoncolburne]

https://github.com/WebOfTrust/keripy/pull/555/files#diff-a4229d5895c922898cdeb03b5bb5c550ea2bf933cef6772027ffa763936ca630R734 here too, though I'm less sure here - you may need a new class? I think if you did, it would be very similar and probably a small refactor

[rodolfomiranda]

looks like it may need mods

Thanks for the catch. I added the change there and in other places as well.

here too, though I'm less sure here - you may need a new class? I think if you did, it would be very similar and probably a small refactor

Changed there too. I don't think a new class is needed. Everything is handled by the same http.clienting.Client with the proper initialization.

[rodolfomiranda]

oops, tests failing. Fixing...

[pfeairheller]

I'm afraid we need a more comprehensive solution to this. Just defaulting to either HTTP or HTTPs may cause problems in the future. Now that we want to support. HTTPs, I think we should allow for it to fail if only HTTP is available which this code will not do.

We probably need to figure out how to make this a parameter passed into the methods for querying or publishing to a witness.

[rodolfomiranda]

I'm afraid we need a more comprehensive solution to this. Just defaulting to either HTTP or HTTPs may cause problems in the future. Now that we want to support. HTTPs, I think we should allow for it to fail if only HTTP is available which this code will not do.

We probably need to figure out how to make this a parameter passed into the methods for querying or publishing to a witness.

isn't the scheme in the OOBI resolution such parameter?

[pfeairheller]

I'm afraid we need a more comprehensive solution to this. Just defaulting to either HTTP or HTTPs may cause problems in the future. Now that we want to support. HTTPs, I think we should allow for it to fail if only HTTP is available which this code will not do.
We probably need to figure out how to make this a parameter passed into the methods for querying or publishing to a witness.

isn't the scheme in the OOBI resolution such parameter?

No. This is about endpoint URLs, not the OOBIs. They are unrelated.

[pfeairheller]

It doesn't matter what scheme is used for an OOBI, the resolution of that OOBI can return endpoints with any scheme they want. The controller may only want to use witnesses that expose HTTPs. With this code, the user could end up using witnesses with HTTP and not even be aware.

[rodolfomiranda]

It doesn't matter what scheme is used for an OOBI, the resolution of that OOBI can return endpoints with any scheme they want. The controller may only want to use witnesses that expose HTTPs. With this code, the user could end up using witnesses with HTTP and not even be aware.

got the point. This patch definitively does not solve that problem. It only allows https as an alternative to http endpoints.

[pfeairheller]

Decision made in last keri-dev community meeting to merge this PR as is and open a new issue for more enhanced support of HTTPs allowing the controller initiating a connection to specify the schemes that they want to use.