SELECT InitDateTable.date_id AS init_date_id
,EndDateTable.date_id AS end_date_id
FROM myschema.mytable MyTable
INNER
JOIN myschema.mydatestable InitDateTable
ON InitDateTable.date = MyTable.init_date
INNER
JOIN myschema.mydatestable EndDateTable
ON EndDateTable.date = MyTable.end_date

You can’t create index because it tells you this view contains a self join, but this isn’t true because it is accesing the same table two times for two separate fields. This is not a SELF JOIN but it tells you it is.

Is it any way to do this?

SELECT InitDateTable.date_id AS init_date_id
,EndDateTable.date_id AS end_date_id
FROM myschema.mytable MyTable
INNER
JOIN myschema.mydatestable InitDateTable
ON InitDateTable.date = MyTable.init_date
INNER
JOIN myschema.mydatestable EndDateTable
ON EndDateTable.date = MyTable.end_date

You can’t create index because it tells you this view contains a self join, but this isn’t true because it is accesing the same table two times for two separate fields. This is not a SELF JOIN but it tells you it is.

Is it any way to do this?

I belive – this limitation will be eliminated in next version of SQL Server, probably – with usage of special join hint (self join executed exactly as inner join).
From architectural point – that happens when attempt to implement generic solution for complex case with far functional dependencies.
From logical point it might be solved as following:
- specify join hint to eliminate the cause of limitation
- give query analyzer more smart to use only INNER join if the view is indexed

Please don’t take it otherwise but we are bored of these limitations. Please change the topic.

Regards,
Amit