You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We need to determine what this specification will have to say about:
Package Registrars
"How does a package developer register a package."
Resolution of package names to package data.
"How does a developer resolve a package name to the released package itself"
I propose that we use EIP137 for this. The benefits that I see are in it's simplicity as well as it already having a degree of community consensus.
For the question of how developers register packages, I suggest that we mark that as being out of scope and not covered in this specification. That doesn't mean we can't come out of the gate with a Registrar that people can use, but I would suggest that the spec should not cover the actual registration process because that is something that each registrar will inherently implement differently.
For the question of how package names are resolved I suggest one or all of the following options.
Use the addr record type that would be expected to resolve to the address of a Package contract. This Package contract would expose some portion of the actual package manifest data via functions, or possible just very basic metadata like released versions and URI's to the actual manifests for those releases.
Define a new uri record type that would allow resolution to arbitrary URI's. This would then allow pointing to an IPFS/SWARM/HTTP uri which would contain the package manifest.
Define a new package record type. (it is unclear to me what data this should return).
The text was updated successfully, but these errors were encountered:
We need to determine what this specification will have to say about:
I propose that we use EIP137 for this. The benefits that I see are in it's simplicity as well as it already having a degree of community consensus.
For the question of how developers register packages, I suggest that we mark that as being out of scope and not covered in this specification. That doesn't mean we can't come out of the gate with a Registrar that people can use, but I would suggest that the spec should not cover the actual registration process because that is something that each registrar will inherently implement differently.
For the question of how package names are resolved I suggest one or all of the following options.
addr
record type that would be expected to resolve to the address of aPackage
contract. ThisPackage
contract would expose some portion of the actual package manifest data via functions, or possible just very basic metadata like released versions and URI's to the actual manifests for those releases.uri
record type that would allow resolution to arbitrary URI's. This would then allow pointing to an IPFS/SWARM/HTTP uri which would contain the package manifest.package
record type. (it is unclear to me what data this should return).The text was updated successfully, but these errors were encountered: