i see this as 2 closely related proposals:

1. map alphanumeric fragment identifiers to elements with the corresponding xml:id attribute in the root document's content.xml

this enables a common use case, users want to reference some heading or bookmark in the document with a URI.

2. map full paths in fragment identifiers to arbitrary elements with the corresponding xml:id attribute in arbitrary files in the ODF package

this is a pretty esoteric use-case, how often do end-users want to reference for example some element inside an embedded object inside the document?

(maybe this was intended to allow for using such URIs with fragment identifiers as a base URI to resolve relative URIs against, but unfortunately that is not possible because RFC 3986 "5.1. Establishing a Base URI" says that fragments are stripped off URIs first.)

so i'm in favor of 1. but not necessarily 2.

As per ODF TC minutes on 21-March-2016, the xml:id in context.xml is accepted for ODF 1.3

Outside of context.xml, a separate JIRA issue should be opened if the feature is to be pursued.

to answer the question what kind of fragment identifiers LibreOffice supports:

• Writer:

• "abc|region" => section named "abc"
• "abc|outline" => heading with number "abc" , or with text that starts with "abc"
• "abc|frame" => text frame named "abc"
• "abc|graphic" => image named "abc"
• "abc|ole" => embedded object named "abc"
• "abc|table" => table named "abc"
• "abc|sequence!1" => 1st sequence field named "abc"
• "abc|text" => search for text content "abc"
• "abc" => bookmark named "abc" or hyperlink named "abc"

• Calc

• anything that can be parsed as a valid cell address or range

• Impress/Draw

• "action?jump=firstslide" => first page
• "action?jump=lastslide" => last page
• "action?jump=nextslide" => next page
• "abc" => page named "abc" or shape named "abc"

• Impress, while slideshow is running:

• "page1" => 1st page
• "abc" => shape named "abc"