bug#81529: 31.0.91; [ELPA] Atom feed entries only specify a self-link

Daniel Mendler <[email protected]> writes:

> Philip Kaludercic <[email protected]> writes:
>
>>> It would be useful to also specify an additional rel="alternate" link
>>> for each feed entry pointing to the package website, e.g.,
>>> https://elpa.gnu.org/packages/tmr.html. This way feed readers like
>>> Elfeed can point users to the package website.
>>
>> As we don't have any structure within the HTML changelog (and we don't
>> reproduce the entire changelog, but both cap it ourselves and only
>> generate what the maintainers include in their news source), we would
>> have to use the same "alternate" link in every entry.
>
> Yes, I am aware of this problem. My suggestion is to add ids to the
> headlines in the html file - this will be useful in any case. Then the
> feed link could also point to the changelog for example:
>
> https://elpa.gnu.org/packages/consult.html#news

Sure, but my point is that it remains the same link for every <entry>.
It is not the end of the world, adjusting the code is trivial, I just
wanted to think this through to not miss any obvious alternative ideas.

>> Do you know if we can just set the alternate link for the entire feed,
>> or will news readers not make use of this?
>
> I don't know if the spec allows this. The entire feed link is always a
> rel="self" afaict. At least Elfeed only supports only rel="alternate"
> links per entry.

The spec implies that it is possible.  [0] says that a feed may have a
link tag, and a comment further below implies that it is possible to
have a rel="alternate" tag as well:

   o  atom:feed elements MUST NOT contain more than one atom:link
      element with a rel attribute value of "alternate" that has the
      same combination of type and hreflang attribute values.

It also makes sense to me, as a HTML page can have alternate link to the
atom feed and the atom feed can have a backlink to the HTML page.

Whether or not clients make use of this and use it as a fallback if an
entry lacks a <link rel="alternate" ...> tag is a different question
that I don't know, but it doesn't seem unreasonable.

[0] https://www.rfc-editor.org/info/rfc4287/#section-4.1.1

>>> Daniel