PEP 694: Update Post-History

peps/pep-0694.rst

@@ -14,6 +14,7 @@ Post-History: `27-Jun-2022 <https://discuss.python.org/t/pep-694-upload-2-0-api-
               `27-Sep-2025 <https://discuss.python.org/t/pep-694-pypi-upload-api-2-0-round-2/101483/31>`__
               `07-Dec-2025 <https://discuss.python.org/t/pep-694-pypi-upload-api-2-0-round-2/101483/35>`__
               `26-Jun-2026 <https://discuss.python.org/t/pep-694-pypi-upload-api-2-0-round-3/107923>`__
+              `29-Jul-2026 <https://discuss.python.org/t/pep-694-pypi-upload-api-2-0-round-4/108320>`__
 
 
 Abstract
@@ -1846,7 +1847,7 @@ as experience is gained operating Upload 2.0.
 Change History
 ==============
 
-* TBD
+* `29-Jul-2026 <https://discuss.python.org/t/pep-694-pypi-upload-api-2-0-round-4/108320>`__
 
   * Add an **Atomic Publication and Conflicts** section.  Specify that publication is atomic with respect to
     the release's filename namespace: the server reserves the session's filenames for the duration of a
@@ -1871,23 +1872,23 @@ Change History
   * Expand the "Why is the project name required" FAQ to note that single-project sessions still improve
     multi-project releases (all projects can be fully staged before a final step publishes each one) and lay a
     foundation for a possible future "publish multiple projects" endpoint.
-  * Define what happens when a publish is requested while file uploads are still in flight, which
-    was previously unspecified.  Publication now has an explicit precondition: every entry in the
-    session's ``files`` mapping **MUST** be in the ``completed`` state, otherwise the server
-    **MUST** reject the publish request with a ``409 Conflict`` identifying the offending file(s)
-    and leave the session editable.  The rule is written as an allow-list so that files in
+  * Define what happens when a session publication is requested while file uploads are still in
+    flight, which was previously unspecified.  Publication now has an explicit precondition: every
+    entry in the session's ``files`` mapping **MUST** be in the ``completed`` state, otherwise the
+    server **MUST** reject the publish request with a ``409 Conflict`` identifying the offending
+    file(s) and leave the session editable.  The rule is written as an allow-list so that files in
     ``error``, and any file states added by future revisions, block publication rather than being
     silently included or dropped.  Add a corresponding row to the publishing session transition
     table and scope the filename reservation to the fully uploaded files the session will publish.
-  * Rename the file upload session ``complete`` status to ``completed``.  This matches the
-    past-participle form of the other settled statuses (``canceled``, and the publishing session's
-    ``published``), and disambiguates the *state* from the ``complete`` *action* and its
-    ``links.complete`` endpoint, which keep their names.
   * Specify that a publishing session **MUST** be cancelable regardless of the states of its files, and in
     particular **MUST NOT** be refused cancellation because a file upload session is ``processing``.  Because
     canceling guarantees nothing will be published, no in-flight validation outcome can affect the result, so
     unlike an individual file deletion there is no race to protect against; the server **MAY** let in-flight
     processing finish and discard the result rather than interrupting it.
+  * Rename the file upload session ``complete`` status to ``completed``.  This matches the
+    past-participle form of the other settled statuses (``canceled``, and the publishing session's
+    ``published``), and disambiguates the *state* from the ``complete`` *action* and its
+    ``links.complete`` endpoint, which keep their names.
   * Remove ``canceled`` from the valid values of the publishing session ``files`` mapping ``status`` key,
     resolving a contradiction with the existing rule that canceling or deleting a file removes its entry from
     that mapping.  The file's own file upload session status URL continues to report ``canceled``.