PEP 503: Specify index content format

[chrahunt]

Currently PEP 503 states

Within a repository, the root URL (/) MUST be a valid HTML5 page with a
single anchor element per project in the repository.

"HTML5 page" is used in a few other places to refer to the expected return value of the PEP 503 repository.

HTML5 specifies two syntaxes/document formats: HTML and XML. Currently pip only supports HTML documents.

I would propose the following:

1. Add a note to PEP 503 to the effect of:

   Below, "HTML page" can be taken to mean an HTML document confirming to [[HTML5]].

   and then replace each instance of "HTML5 page" with "HTML page".

2. Add a constraint to the existing list

   When served via HTTP, repositories SHOULD include a Content-Type header with a media-type value of text/html when returning an HTML page. Clients MAY choose to parse an HTTP response body as an HTML page in the absence of a Content-Type header.

References

1. Fails with PEP 503-compliant XML-serialised HTML5 ("XHTML5") directory indices pypa/pip#6697
2. https://html.spec.whatwg.org/multipage/introduction.html#html-vs-xhtml

[brettcannon]

@dstufft

[KloolK]

I'm trying to use the auto-indexed web version of an SVN repository as a repository for pip. Unfortunately, the pages are served with Content-Type: text/xml and therefore pip currently skips the repository. I thought I'd add this here as a real-world example why this could be useful.

[hugovk]

We try to avoid discussing the content of PEPs themselves in this repo, as there's not many people following here.

Instead, use the place listed under the Pep's "Discussions-To" header.

In this case, it's the now-archived Distutils-SIG list, so please post to the Packaging category at https://discuss.python.org/c/packaging/14