* chore: Unpin django-stubs / update mypy
* test: Use "app.model" for ManyToManyField
django-stubs requires "app.model" instead of just "model" for
ManyToManyField lazy model references.
See https://github.com/typeddjango/django-stubs/issues/1802
* fix: Properly set AD as action holder when submitting to IESG for publication (#5227)
The clear intent of `to_iesg` is that
a) the document AD should be the group AD, if not already set, and
b) the document Action Holder should be the document AD;
but there was an order-of-operation error,
such that the Action Holder remained empty.
* refactor: Don't take values out of the parent scope, don't insert values into the parent scope
* refactor: Streamline DocEvent creation
* fix: Wrap weasyprint to catch exceptions (#6324)
* test: Restore socket function after test
* test: Use mock instead of monkeying with sockets
* refactor: Log the error
* fix: Don't catch non-Exception interruptions
---------
Co-authored-by: Robert Sparks <rjsparks@nostrum.com>
* feat: Allow entering agenda text directly (#6532)
* fix: Hide label as well as file/text input box
* refactor: Package javascript for static/dist
* fix: Fix test cases broken by view changes
* test: Add test case for entering agenda text
* refactor: assertRedirects
* fix: don't create a volunteer object when a person has already volunteered
* fix: safer create
* refactor: use shorthand from django
* fix: also protect volunteer from datatracker from race
* fix: clean up shadowed name in document_referenced_by.html
* fix: include refs to rfc's came_from_draft()
* fix: include refs to draft's became_rfc()
* fix: Count indirect refs by RFCs
* refactor: break indirect ref_by counting to its own fn
* fix: only count refs to pre-rfc draft, not post-draft rfc
(and rename a method)
* test: test referenced_by_rfcs methods
The test_referenced_by_rfcs_as_rfc_or_draft() test
fails because there's a bug!
* test: actually, do double-count refs to rfc/draft
Let's do include refs to an rfc and its precursor draft
as separate refs. This almost surely indicates a data
error because it would mean an rfc referenced both an
rfc and the draft that it came from. That should never
be allowed, so at least let some light fall on it if
it happens.
* chore: Add docstring to document_referenced_by view
* Fix AD queue graphs and AD doc list sorting.
Fixes#6700Fixes#6720
* Fix the issue @rdanyliw found. Hopefully.
* More fixes
---------
Co-authored-by: Robert Sparks <rjsparks@nostrum.com>
* feat: Capture volunteers from the registration system. Fixes#5938
* fix: allow withdrawn field to be null
* docs: add explanatory code comments
* fix: identify current nomcom by is_accepting_volunteers
* fix: use auto_now_add instead of auto_now
* fix: update migration
---------
Co-authored-by: Robert Sparks <rjsparks@nostrum.com>
* fix: Avoid accessing slug when state is None
* fix: ignore pre-rfc drafts as rfc-to-rfc references
* fix: also ignore pre-rfc to pre-rfc refs
* chore: remove breakpoint
* fix: ignore subseries non-downrefs
* fix: label nodes as RFCS when possible
* fix: get "rfc" flag right
* chore: Remove comment, answer seems to be "no"
* refactor: Specify state type; drop redundant source doc type check
We're only admitting source__type="draft", so
no need to check it again in filter queries
* refactor: Specify state type in old code
* feat: Sort RFCs on subseries doc pages
* feat: Sort RFCs on subseries index pages
* fix: Show subseries doc name as page title
* refactor: Use prettystdname filter for consistency
* fix: Combine Group and GroupHistory when counting
* refactor: Use same symbol in parallel iterators
* fix: Handle group acronym changes in assertion
* refactor: Avoid reference to GroupHistory in assertions
* chore: Revert so assertion fails on changed acronym
The filtering and meeting displays disagree on how to handle this,
so let's get a notification if it comes up in production.
