Ed: I dont understand the bug ... Is this a problem with the parse output from Parsoid or is the problem with how Parsoid serializes edited HTML in VE? Can you clarify and provide wikitext snippet, if the former?

It's a problem with serialization. If we provide two adjacent references that are clones of existing references (via copy/paste) and both don't have an about attribute, Parsoid seemingly considers them both to be in the same about group with about=undefined and drops all but the first.

Minor nitpick: it would be null. But otherwise Roan's hunch sounds very plausible to me.

This (or a closely-related bug) are really badly breaking normal editing in VE, which we thought we had fixed with bug 52228 - namely, making multiple sequential references.

Minimum spanning test:

Input HTML+RDFa:

<p>Foo<span typeof="mw:Extension/ref" data-mw="{&quot;name&quot;:&quot;ref&quot;,&quot;body&quot;:{&quot;html&quot;:&quot;Bar&quot;},&quot;attrs&quot;:{}}"><a href="#cite_note-1">[1]</a></span><span typeof="mw:Extension/ref" data-mw="{&quot;name&quot;:&quot;ref&quot;,&quot;body&quot;:{&quot;html&quot;:&quot;Baz&quot;},&quot;attrs&quot;:{}}"><a href="#cite_note-1">[1]</a></span></p>

Expected WT:

Foo<ref>Bar</ref><ref>Baz</ref>

Actual WT:

Foo<ref>Bar</ref>

Change 81886 had a related patch set uploaded by GWicke:
Bug 53434: Only skip about siblings if about is actually set

https://gerrit.wikimedia.org/r/81886

Change 81886 merged by Arlolra:
Bug 53434: Only skip about siblings if about is actually set

https://gerrit.wikimedia.org/r/81886

• Bug 53334 has been marked as a duplicate of this bug. ***

Not sure if has been rt-tested and deployed yet.

Deployed to production this morning and verified fix.