Hi All,
(I'm new to HDF so this may be a noobie question).
I'm working over large datasets, ordered as datatables. The data
is date-time ordered what makes easier to find the datalines I want. After
find the datalines I'm selecting those lines with hyperslabs (line by line
to work over them), but I've found a situation where would be really useful
to have a selection of the whole lines I want (with an hyperslab) and the
get the dataspaces line by line from that zone selection. Is there any way
to do that without creating a temporary dataset with the big selection and
work over it there?
(a single note: i'm not working with any higher level API, I'm just
including hdf5.h from HDF5 1.8.9 version)
ยทยทยท
____
In case i'm not explaining myself properly above, here goes a simplified
example:
[1]- I have some data lines
_______________________
_____|______|_____|_____|
_____|______|_____|_____| <<< (select line)
_____|______|_____|_____| <<< (select line)
_____|______|_____|_____| <<< (select line)
_____|______|_____|_____|
[2]- I want to update data in the lines marked with "<<<", so i'm selecting
that datalines with hyperslabs (in the real case there would be not 3
lines, but far more lines, so I'm selecting the space with an hyperslab,
read the current line, update the data I want, and then overwrite the slab
space with the new data):
__a__|___b__|__c__|__d__| >> read to struct data >> operate over it
(update b to g, c to h ) >> overwrite the slab selected space
>__a__|__g___|__h__|__d__|
[3]- But in a situation here would be really useful to select all the zone
I want:
_____|______|_____|_____| |
_____|______|_____|_____| |<<< (select zone)
_____|______|_____|_____| |
and then iterate it, performing the action described in [2] for each line.
But to do that I would have to get a "line selection" from the "zone
selection" and then tell the zone space I have new data in certain lines...
so here is the point I'm not being able to do, but it would easy once I was
able to create a "selectable" space working as a virtual dataset but only
with the datalines I need.
my thanks,
Saulo