The newest packages feed does not always return the latest packages

[behnazh-w]

The newest packages feed might have timezone issues, which can result in returning old packages instead of the newest ones. For example, at 2024-07-03-13-31+10 I got the following result from the feed: 2024-07-03-13-31+10.zip
containing this example timestamp for publication:

<pubDate>Wed, 03 Jul 2024 01:49:28 GMT</pubDate>

Later I got this feed at 2024-07-04-06-51+10: 2024-07-04-06-51+10.zip, which contained more recent packages as expected. Here is an example timestamp from this feed:

<pubDate>Wed, 03 Jul 2024 20:47:58 GMT</pubDate>

But then I got the following result at 2024-07-04-11-32+10: 2024-07-04-11-32+10.zip that contains packages that are older than the ones obtained previously at 2024-07-04-06-51+10. Here is an example publication timestamp:

<pubDate>Wed, 03 Jul 2024 01:30:29 GMT</pubDate>

So, looks like the most recent feed contains packages that were not really the most recent.

[di]

Hi, this is not a timezone issue and is expected. This could be the case if multiple recent projects were deleted, because our feed just provides the 40 most recently packages at a given time.

warehouse/warehouse/rss/views.py

Lines 87 to 101 in 15de759

	def rss_packages(request):
	request.response.content_type = "text/xml"
	newest_projects = (
	request.db.query(Project)
	.options(joinedload(Project.releases, innerjoin=True))
	.order_by(Project.created.desc().nulls_last())
	.limit(40)
	.all()
	)
	project_authors = [
	_format_author(project.releases[0]) for project in newest_projects
	]
	return {"newest_projects": tuple(zip(newest_projects, project_authors))}

Note that similarly this feed may be missing projects if many new projects were created between multiple requests.

Depending on what you're doing, you may want to use the XML-RPC API or the Simple API instead: https://warehouse.pypa.io/api-reference/feeds.html#project-and-release-activity-details