Another option is to format a gigantic INSERT ... VALUES statement, like
this:
DATA =: 3 3 $ ;:'alpha beta gamma delta episilon zeta eta theta iota'
'INSERT INTO someTable (columnA, columnB, columnC) VALUES ',}: ; <@:(1
A.'(',,&')')@:;"1|:('"',,&'",')&.>DATA
INSERT INTO someTable (columnA, columnB, columnC) VALUES
("alpha","delta","eta"),("beta","episilon","theta"),("gamma","zeta","iota")
Of course, you have to make sure you quote values of different types
(string, char, date, etc) appropriately.
Having said that, I'm not sure this will be much faster than your current
row-at-a-time insertions. For large data sets, the J formatting, network
transmission, and SQL parsing may overcome any gains from bulk insertion.
Also, you may hit edge conditions on the length of a SQL script your server
will accept. The speed of J formatting we can control to an extent; so if
your expression is too slow, post it, and we can improve it. And if the
server rejects your SQL on length, you could try inserting newlines into the
statement to see if that fixes it.
Anyway, for practical matters, there's usually more than one way to skin a
cat. But now I'd like to return to my preferred domain: wholly impractical
matters.
-Dan
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
