Skip to content
Ferrocat Docs

0008: Normalize Headerless PO Files on Write

Status

Accepted

Context

Some upstream PO parser/writer libraries preserve catalogs that omit the header entry entirely:

#, fuzzy
msgid "Save"
msgstr "Speichern"

ferrocat-po currently serializes such catalogs with an explicit empty header:

msgid ""
msgstr ""

#, fuzzy
msgid "Save"
msgstr "Speichern"

This differs from byte-preserving roundtrip behavior, but it reflects the canonical PO shape used by most gettext-oriented tooling and makes later header population straightforward.

Decision

ferrocat-po will normalize headerless PO files on write.

Specifically:

  • parsing headerless PO input remains supported
  • serializing a PoFile without headers emits an explicit empty header entry
  • existing headers are preserved rather than regenerated

This is intentional normalization behavior, not a compatibility bug.

Consequences

  • roundtrips for headerless PO files are not text-identical
  • output becomes more uniform and easier to enrich with metadata later
  • source-attributed conformance cases that expect headerless write-back should be treated as intentional divergence rather than known_gap

Rejected Alternatives

  • Preserve headerless files exactly on write.

This would improve diff stability for that narrow case, but it would keep a less canonical PO shape and make ferrocat-po less opinionated about output normalization.