The colon ':', whether used as a Separator Character or otherwise MUST NOT be percent-encoded

[iseki0]

• The following characters MUST NOT be percent-encoded:
  • the colon ':', whether used as a Separator Character or otherwise, and

purl-spec/PURL-SPECIFICATION.rst

Lines 279 to 285 in c53ba0e

	- The following characters MUST NOT be percent-encoded:
	- the Alphanumeric Characters,
	- the Punctuation Characters,
	- the Separator Characters when being used as ``purl`` separators,
	- the colon ':', whether used as a Separator Character or otherwise, and
	- the percent sign '%' when used to represent a percent-encoded character.

purl-spec/tests/types/oci-test.json

Lines 174 to 191 in c53ba0e

	{
	"description": "Build test for <class 'type'> PURL",
	"test_group": "base",
	"test_type": "build",
	"input": {
	"type": "oci",
	"namespace": null,
	"name": "hello-wasm",
	"version": "sha256:244fd47e07d10",
	"qualifiers": {
	"tag": "v1"
	},
	"subpath": null
	},
	"expected_output": "pkg:oci/hello-wasm@sha256%3A244fd47e07d10?tag=v1",
	"expected_failure": false,
	"expected_failure_reason": null
	}

"expected_output": "pkg:oci/hello-wasm@sha256%3A244fd47e07d10?tag=v1", ?

Maybe the whole testcases should be reviewed?

[iseki0]

Personally, I think we'd better following the RFC, instead of modification the standards manually. Especiallly when we haven't enought force to make everything.

[matt-phylum]

You mean one of the URL RFCs? The RFC rules are more complicated than the PURL rules. Different parts have different sets of characters that must be encoded and PURLs have different parts that URLs in ways that are important to percent encoding. URLs have one path, but PURL has a namespace and a name, and the encoding is different between the namespace and name. URLs have one qualifier string, but PURL has a collection of qualifier keys and values.

Because URL uses different rules for different parts of the URL, and because there are multiple different URL specs, it was common for different PURL implementations to produce different "canonical" PURLs. There was a lot of time spent on encoding rules which eventually resulted in this set of rules.

#438

[iseki0]

@matt-phylum the URI RFC, RFC 3986. In fact the PURL is an URI. So why we define our own escape rules?