On 24 Sep 2014, at 11:53am, Prakash Premkumar <prakash.p...@gmail.com> wrote:
> Thanks a lot Hick,for your approach. > With the approach you suggested, we are creating extra queries and if the > join is on n tables there will be n+1 queries , and each query will have to > go through a query planning stage. > Is there an alternative idea ? Execute the SELECT with the JOIN you originally described and pick the results apart in your programming language instead of importing each row of the result into a separate object. In other words, write your program to do this piece of analysis you described: > When a new row comes in and if the object for that row is already created. > i.e if r11 again comes as output, I will not create a new object,instead I > would use the old object and set pointers between the old T1 and object and > the T2 object (if r21 is new,else do not set pointers and do not create > objects) Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users