# developer.multiDataSetConventions

Ed West and Jeremy were talking about how conventions could be introduced to support reading in multiple datasets in one pass. For example, with an ascii file you want to read in Y(T). This is common, and we support doing this in one pass with DEPEND_0 keywords. But what if you also want to read in DEPEND_1? I've added keywords for this too. Now someone wants to do similar for the binary reader, but can't until I implement it. What if instead there were conventions for how these are specified, and they can be specified uniformly?

x= vap+bin:/tmp/foo.bin?recLength=16&recOffset=4
y= vap+bin:/tmp/foo.bin?recLength=16&recOffset=8
plot( x,y )


You would have:

ds= vap+bin:/tmp/foo.bin?recLength=16&dep0.recOffset=4&recOffset=8
plot( ds )


This would implicitly create a DEPEND_0 dataset read in with the same arguments, except for recOffset. Or more importantly:

x= vap+bin:/tmp/foo.bin?recLength=16&recOffset=4&byteOffset=16
z= vap+bin:/tmp/foo.bin?recLength=16&recOffset=8&byteOffset=16
y= vap+bin:/tmp/foo.bin?recLength=16&recOffset=8&byteOffset=0&recCount=1
plot( x,y,z )


becomes:

ds= vap+bin:/tmp/foo.bin?recLength=16&dep0.recOffset=4&recOffset=8&byteOffset=16&dep1.recCount=1&dep1.byteOffset=0
plot( x,y,z )