Make Packed Data section (CF 8.1) integer types consistent with Data Types section (CF 2.2)

[cofinoa]

Before submitting an issue be sure you have read and understand the github contributing guidelines: https://github.com/cf-convention/cf-conventions/blob/main/CONTRIBUTING.md and the rules for CF changes: http://cfconventions.org/rules.html

Change proposals should include the following information as applicable.
If important to the issue, the problem text should be pasted in the body of the issue and proposed fix included.
A link to the line where the problem exists could also be included.
Alternatively, a pull request can be used to describe the problem and the proposed solution more precisely.

Title

Make CF 8.1 (Packed Data) integer types consistent with Data Types (CF 2.2)

Moderator

@user

Requirement Summary

Make consistent packed data section

Technical Proposal Summary

Make Data Packed section with any-integer in Data Types section

Benefits

Use of the new netCDF4 integer types as storage for packed data (CF-1.9)

Status Quo

See cf-convention/discuss#202

Associated pull request

#425

Detailed Proposal

The Data Packed section (CF 8.1) enumerates integer data types:

... the variable containing the packed data must be of type byte, short or int.

but it misses new integer data types introduced in CF-1.9 (#243 #244 #294).

The proposed change is to remove the explicit enumeration on integer types and add a reference to the Data Types section (CF 2.2):

... the variable containing the packed data must be of an integer type (see <<data-types>>).

[JonathanGregory]

Dear Antonio @cofinoa

Thanks for opening this. Is it related to issue 374, raised by Ken Mankoff?

Best wishes

Jonathan

[cofinoa]

Dear Jonathan (@JonathanGregory),

The reason for this issue was cf-convention/discuss#202. I wasn't aware about issue #374 but after reading it, this issue is also related to it.

The issue #374 is missing all unsigned integers introduce on CF-1.9 that can be also used for data packing.

My proposal would be remove the explicit enumeration of integer types, and make reference to any integer type or an integer type as it is been mentioned on Data Types section (CF 2.2), to fix (IMO) the missing unsigned integer types on Data Pack section (CF 8.1).

With respect to issue #374, I would just keep separated from this and wait for a decision been made with respect to it, and modify (or not) Data Pack section (CF 8.1).

[JonathanGregory]

Dear Antonio @cofinoa

Thanks for looking at it. In that case, we have more urgency to to decide about 374. As you will see from my contribution to that issue, I'm not clear what is intended by the existing text giving restrictions on packing. No-one else has offered an interpretation. It would be very helpful if you or anyone else who uses packing could contribute thoughts to 374.

Best wishes and thanks

Jonathan

[JonathanGregory]

Dear @cofinoa and others

I have proposed text in 374 to resolve this issue as well as that one. Please have a look.

Thanks and best wishes

Jonathan

[JonathanGregory]

I'm closing this issue because I believe it's being addressed in #374. If you disagree, please either reopen this issue, or comment in #374.