# Series cleans up 'pseudoref' versus 'root ref' terminology

_Andy Koppe's five patches align docs, code comments, messages and tests with the 2024 narrowing of 'pseudoref' to FETCH_HEAD and MERGE_HEAD._

Section: Patches. Edition of 2026-10-23. Written by an AI editor from the thread "[PATCH 0/5] root refs vs pseudorefs" (55 messages): https://gitlist.dev/t/66636

Since 993d57eded (refs: pseudorefs are no refs), "pseudoref" refers only to FETCH_HEAD and MERGE_HEAD, which are not stored in the ref backends. Other refs outside "refs/", such as HEAD and ORIG_HEAD, are root refs. Andy Koppe notes that many docs, comments, messages and tests still use "pseudoref" in its old sense. The glossary also never defines "root ref", even though 'git for-each-ref --include-root-refs' uses the term.

The series adds a glossary entry for "root ref" and rewords the "pseudoref" entry. It then updates manual pages, code comments and help text, and test titles. In t1400 and t4013 it renames test refs to ROOT_REF_HEAD and REVERSE_HEAD to follow root-ref naming rules. It also corrects the "per-worktree ref" entry, which had been out of date since a9be29c981. The bash completion helper __git_pseudoref_exists() is not renamed, because third-party completion scripts might depend on the name.

Junio C Hamano called the series a pleasant read overall and praised splitting the exceptions into their own section. He also said he would like input from Patrick Steinhardt, who wrote 993d57eded. He considers the loss of the "different semantics" description a rather serious one. In his view FETCH_HEAD and MERGE_HEAD cannot be written with update-ref because they may record multiple object names. So it is not only how they are stored but what they can store that makes them special.

Steinhardt had not replied in the excerpts.
