On Tue, Jun 21, 2011 at 4:38 PM, Johann Hibschman <[email protected]> wrote: > Raul Miller <[email protected]> writes: > >> For starters, I would parse your data like this: >> >> 'area source year qtr hpi'=:|:}.<;._2@,&',';._2 htData >> 'year qtr hpi'=:".L:0 year;qtr;hpi > > Thanks, that's interesting. It took me a depressingly long time to > understand that first phrase, but it was illuminating. I'd not used > cut at all before. For the second, I'll have to work on understanding > L: more. My first impulse would be to use something like ".@< . I > still don't quite understand when L: is appropriate.
each (or &.>) would also work here. >> That said, I think tables in J are best represented as columns -- each >> column has a name (it's a variable) and columns from the same table >> are the same length (indices on one column work on the other columns >> from that table). > > Well, I was thinking that if I had the entire table stored as a boxed > array of some sort, then I could do selections on the entire table at > once. For example, when I was trying to extract just the subset that > had area-.'US', if I could do You could do that but then you have to deal with extracting columns every time you use it. A better idea, I think, if you want to tightly associate the columns, would be to make them all names in the same locale. But if you are only working with a small number of tables, I think that bundling them is more work than it is worth. -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
