Not sure this helps, a way to a conditionally insert based on if record already exists, is a select with literals left outer joined to the maybe record and use a where test value is null.
Something like this pseudo SQL insert into T (valueA, valueB') (select 'ValueA', 'ValueB' left outer join T where T.valueA ='valueA' and T.valueA is null); On 11/18/2019 2:14 PM, Simon Slavin wrote: > On 18 Nov 2019, at 10:00pm, Jose Isaias Cabrera <jic...@outlook.com> wrote: > >> Thanks Keith. So, you are saying that this is a bad INSERT, and I don't >> know much to argue, but is working. If I take out the first IfNull, and >> there is not, at least one instance of 'p006' in the table, the INSERT never >> works. I was thinking of using COALESCE, but that would also mean that one >> or the other would have to be not null. Any suggestion would be appreciated. > Being completely serious, whenever I see "undocumented" or "implementation > dependent" or "optimization side-effect", or a SQL statement I can't parse in > my head, I usually decide to do it in my programming language instead. This > simplifies testing and debugging, and makes things easier for the poor > engineer who has to understand my code. > > You can do clever things in a language like SQL which allows recursive > construction clauses. But what strikes me as ingenious when I'm writing it > can look bizarre and impenetrable to me, or someone else, a year later. > _______________________________________________ > sqlite-users mailing list > sqlite-users@mailinglists.sqlite.org > http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users > _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
