Fix for issue #6365

[vchoi-hdfgroup]

Clarify the description for the chunked dataset's dimensions.

[jhendersonHDF]

Shouldn't these stay 0-based and instead "Dimension #n" below should be changed to n - 1? Otherwise, this seems to conflict with the text below that uses "(n - 1)" when referring to dimensions stored.

[vchoi-hdfgroup]

If you look at the field "dimensionality" description, "For chunked storage, the value stored is one greater than the dataset’s actual number of dimemsions. For example, a 1-dimensional dataset stores a value of 2. This increment accounts for the dataset element size which occupies the final dimension entry."

[jhendersonHDF]

But assuming indices are still 0-based here, what's stored is the sizes for dimension 0..(n-1) and the dataset element size just happens to be an extra element in that "array".

[vchoi-hdfgroup]

When you mentioned above conflicting with the text "(n-1)" which line are you referring to?

[jhendersonHDF]

In this diff it's line 7700. That text pretty much states what I'd expect: For an n-dimensional dataset (for n >= 2), dimensions 0..(n-1) are stored and then the dataset element size is stored. The dataset element size is part of the dimensions "array", but it's not really a dimension itself.

[vchoi-hdfgroup]

So actually I should have the following:
In the layout table:
"Dimension 0 Size, Dimension 1 size,.... Dimension (n-1) size, Dataset element size"

And in the description table:

For chunked storage, the dimensions define the size of a single chunk. They are in units of array elements (not bytes). The first dimension stored in the list of dimensions is the slowest changing dimension and the (n - 1) dimension stored is the fastest changing dimension.
For an n-dimensional dataset (for n >= 2), dimensions 0..(n-1) are stored and then the dataset element size is stored. The dataset element size is part of the dimensions "array", but it's not really a dimension itself.

[jhendersonHDF]

"Dimension 0 Size, Dimension 1 size,.... Dimension (n-1) size, Dataset element size"

Yes, this part is what I think the layout table should say.

For an n-dimensional dataset (for n >= 2), dimensions 0..(n-1) are stored and then the dataset element size is stored. The dataset element size is part of the dimensions "array", but it's not really a dimension itself.

These were just my own words summarizing the encoding; I think the text currently in the description table in this diff is fine as is and you don't need to add these words.

[github-actions]

Review Checklist

This PR touches the following areas. Each needs a sign-off
from its listed owners before merging.

• docs — @brtnfld, @gheber, @glennsong09, @lrknox