Evandro's mailing lists (Please, don't send personal messages to this
address) schrieb:
Yes it is exactly that. I will follow you advice and create a
abstraction layer for the data access that will return the sparse
dataset using the standard dataset as input.
There is just one thing I disagree you said it that the performance is
not good, right. However, it is practical! Nothing is easier and more
practical than keeping the sparse representation inside of the database
for my application.
I bet even your application would profit from not handling sparse data.
You could just "not" insert them into your tree instead of having
to jump over empty elements.
And there is always a way to lazily abstract the data to some
frontend (While I doubt anybody can actuall scroll wide enough on
a screen to see all the 1600 colums ;)
---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match
