Paul Wessel

Feedback from gs developers. They had a look at the initial PS file. We all agree it is not a bug. However, their comment gives useful information back to us...

Sure, we wrote **grdcontour** back in the day (1988?) so not much info to go by - we invented our own algo. Of course it should be rewritten by anyone...

Basically, I am setting the 3-d dimension to chunk = 0. I think the order of the _nc4_chunksize_ array is such that the first item is z, then y and...

It is not just chunking at work. There is IO_NC4_DEFLATION_LEVEL too - how do those differ?

As for letting IO_NC4_CHUNK_SIZE, since we have been doing y,x chunking syntax for years, should we do this: If 2 values, then assume y,x If 3 values, then assume z,y,x...

The benefit of the latter is that we dont have to change the setting when moving from 2-D to 3-D entities.

Yes, I guess it is OK. If we find three it is z,y,x and if someone writes a grid we use y,x only. So should be safe and the docs...

Update: Not sure if chunking in z will be possible. Right now, GMT writes the cube as a series of layers. Each layer have padding so we using the _gmtnc_io_nc_grid_...

We dont really have any thing that is truly 3-D in GMT so this is a kludge via stacks of grids, hence each grid has a pad, etc. have a...

Also, while the documentation says one of **+v** or **+w** can be set, the code is happy to accept both. From the logic it seems acceptable to set both but...