IssueTracker

In principle, BigQuery seems to allow multiple partition windows (of different scope):

select
s.word
,sum(s2.word_count) over (partition by s.corpus) as count_corpus
,sum(s2.word_count) over () as count_total

from publicdata:samples.shakespeare s
inner join publicdata:samples.shakespeare s2
on s2.word = s.word
and s2.corpus = s.corpus

However, when I try to do this with our own tables, I am running into an unhelpful error message to the effect that the first field in the select statement cannot be found.

Job ID: snap-sale:job_SPeZfcVqIHlB0fzrsnAr7kcbhnY

I don't see any structural difference between these two queries, except that in the first query I am joining the table with itself.

Could you find out what causes this error, and if it cannot be fixed right away, change the error message into something less misleading?