我不确定以前是否有人问过这个问题，但我在不同的论坛上搜索过，没有找到确切的答案。我们是否应该总是将单词RECURSIVE初始化为第一个CTE（即使它不是递归CTE），以便整个查询运行？
示例查询：

-- initalising some data
with temp_cte1 as (
  select
  100 as id,
  'test' as name,
  30000 as salary,
  date('2011-10-28') as start_date,
  date('2011-11-30') as end_date
),

-- this cte is to get all the dates between the start and end dates
temp_cte2 as (
  select
    id,
    start_date as curr_day,
    end_date
  from temp_cte1
  union all
  select 
    id,
    DATE_ADD(curr_day, INTERVAL 1 DAY) as curr_day,
    end_date
  from temp_cte2
  where curr_day < end_date
),

-- this cte is filter out specific dates from tempcte2
temp_cte3 as (
  select
    id,
    count(*) as pay_days
  from temp_cte2
  where DAY(curr_day) in (1, 15)
  group by id
)

select 
  d.id,
  d.name,
  d.salary,
  d.start_date,
  d.end_date,
  f.pay_days
from temp_cte1 d
left join temp_cte3 f
on d.id = f.id

上面的查询抛出表temp_cte2不存在的错误：错误响应：错误代码：1146.表'final.temp_cte2'不存在

-- initalising some data
with recursive temp_cte1 as (
  select
  100 as id,
  'test' as name,
  30000 as salary,
  date('2011-10-28') as start_date,
  date('2011-11-30') as end_date
),

-- this cte is to get all the dates between the start and end dates
temp_cte2 as (
  select
    id,
    start_date as curr_day,
    end_date
  from temp_cte1
  union all
  select 
    id,
    DATE_ADD(curr_day, INTERVAL 1 DAY) as curr_day,
    end_date
  from temp_cte2
  where curr_day < end_date
),

-- this cte is filter out specific dates from tempcte2
temp_cte3 as (
  select
    id,
    count(*) as pay_days
  from temp_cte2
  where DAY(curr_day) in (1, 15)
  group by id
)

select 
  d.id,
  d.name,
  d.salary,
  d.start_date,
  d.end_date,
  f.pay_days
from temp_cte1 d
left join temp_cte3 f
on d.id = f.id

但是如果在起始CTE中使用关键字RECURSIVE进行初始化（即使CTE不是递归的），查询也可以正常运行。有人能解释一下为什么我应该在顶部使用关键字RECURSIVE初始化递归CTE吗？
我想知道为什么我们必须在第一个cte中使用关键字RECURSIVE。