id	summary	reporter	owner	description	type	status	component	version	severity	resolution	keywords	cc	stage	has_patch	needs_docs	needs_tests	needs_better_patch	easy	ui_ux
10962	GROUP BY generation and extra seems to be too greedy	Brian Rosner	nobody	"I may not understand the ORM code fully, but this seems like wrong behavior. Take the following queryset:

{{{
queryset = channel.message_set.all().values(
   ""nickname"",
).annotate(
   message_count = models.Count(""id""),
).extra(select={
    ""percentage"": ""FLOOR((COUNT(*) / %d.0) * 100)"" % channel.message_set.count(),
}).order_by(""-message_count"")
}}}

I am seeing the following query generated:

{{{
SELECT (FLOOR((COUNT(*) / 1105908.0) * 100)) AS ""percentage"",
       ""irc_message"".""nickname"", COUNT(""irc_message"".""id"") AS ""message_count""
FROM ""irc_message""
WHERE ""irc_message"".""channel_id"" = 3
GROUP BY ""irc_message"".""nickname"",
         FLOOR((COUNT(*) / 1105908.0) * 100)
ORDER BY message_count DESC LIMIT 10
}}}

As you can see the expression from extra(select=) is sneaking into the `GROUP BY` clause. Tracking this down led to #10132 which seems to introduce this new behavior (even though I don't know if things worked before [9838])."		closed	Database layer (models, ORM)	dev		invalid			Unreviewed	0	0	0	0	0	0