Feature Request: add sanity checks to hsw-cli when adding/modifying records

[mrose17]

Some of the data in the HNS root zone are "wrong", e.g., IP addresses being used as the name of NS RRs (instead of as the value of the RR).

i suggest that hsw-cli be extended to include a module that does a syntax/sanity check when records are created/modified, e.g., as in:

    hsw-cli rpc sendupdate easyhandshake '{"records": [ ... ]}'

If you go to any DNS service that let's you edit RRs, they all perform such checks (or at least they should). I suggest taht hsw-cli do the same!

[tynes]

There is an RPC on the node that can be used for this, validateresource
See handshake-org/hsd#429

[pinheadmz]

@mrose17 thanks for bringing this up. hs-client is just an empty wrapper client to make calls to an hsd node, so features like this would likely go in hsd itself. Like @tynes said we have a validateresource rpc already that checks for basic syntax errors in the resource JSON object before the user calls sendupdate -- but perhaps it could be expanded.

For example, reminiscent of Bitcoin Core's testmempoolaccept, validateresource could return a complex JSON object listing arrays of "errors" and "warnings", where errors are the current syntax tests and warnings are violations of traditional DNS configuration.

Possible example result

{
  errors: [
    "Invalid DS record. KeyTag must be a uint16.",
    "Invalid GLUE6 record. ns must be a valid name."
  ],
  warnings: [
    "Unknown DS algorithm identifier",
    "DS Digest length does not match digestType",
    "Glue record is not a subdomain of the zone"
  ]
}

Marshall I'd appreciate your suggestions for things to test

[mrose17]

@pinheadmz - when in doubt, see what the isc did with bind. in this case, we have https://kb.isc.org/docs/aa-01127 ... so my short answer is to get the source for that and integrate that into validateresource. hopefully, you can just copy-paste without having to do any transcoding by hand.

the other thing I would check is to see how the invalid RRs got into the HNS zone. if sendupdate wasn't used, i suppose there isn't much that can be done; otherwise, it is almost certainly the case that when updating records that a syntax check is mandated. (although I suppose you could add a "i know what i'm doing and please do not sanity-check this parameter" option on sendupdate).

does that make sense?