Clarification of flag_meanings attribute

[martin.juckes]

1. Title

Clarification of flag_meanings attribute

2. Moderator

Jonathan Gregory

3. Requirement

Clear definition and greater range of characters in flag_meanings attribute.

4. Initial Statement of Technical Proposal

Section 3.5 of the convention states "The flag_meanings attribute is a string whose value is a blank separated list of descriptive words or phrases, one for each flag value.", and a "word" is currently interpreted by the CF-checker to consist only of alphanumeric characters, ruling out hyphens and brackets which may be useful in this context. The proposal is that a word in this context should be defined to consist of alphanumeric characters plus underscore '_', period '.', plus '+', hyphen '-', or "at" sign '@', to be consistent with the character set allowed for netCDF variable/dim/attr names [there is no particular need for this consistency, except to avoid having multiple definitions of what constitutes a word].

The above sentence from section 3.5 should be modified to read "The flag_meanings attribute is a string whose value is a blank separated list of descriptive words or phrases, one for each flag value. Each word or phrase consisting of characters from the alphanumeric set or from the following four: '-', '.', '+', '@'.",

5. Benefits

More meaningful "flag_meanings" will be possible.

E.g. for a dataset of eco-regions, flag_meanings such as "Alberta-British Columbia foothills forests" could be used.

6. Status Quo

There is ambiguity in the current standard, which I believe should be alleviated.

[jonathan]

Dear Martin

Thanks for the proposal. I've got a couple of remarks:

• Underscore is also allowed, isn't it? If so, the text could say. "Each word or phrase should consist of characters from the alphanumeric set and the following five: '_', '-', '.', '+', '@'."

• "Alberta-British Columbia foothills forests" would still not be a possible word or phrase, because it has spaces in it. That, I think, was one the original points at issue. When this part of the convention was added, I think it was expected that spaces would be replaced by hyphens in a phrase like that.

Also, I note that the CF conformance document for section 3.5 should be altered in a corresponding way.

Cheers

Jonathan

[martin.juckes]

Dear Jonathan,

Your are right all points: the example I gave should have been: "Alberta-British_Columbia_foothills_forests".

cheers, Martin

[jonathan]

Replying to martin.juckes:

OK, fine, thanks.

In that case, I invite discussion on this proposal. As it's a straightforward and limited one, which has already been discussed on the email list, I don't expect this will be a complicated discussion.

Jonathan

[apamment]

Dear Jonathan and Martin,

I have read the discussion in this ticket and the preceding one on the mailing list and I support the proposal to extend the character set permitted in the flag_meanings attribute. The additional characters would enhance the readability of the descriptive strings and make them more easily understandable.

Alison

[jonathan]

Sufficient support has been expressed and no objections raised to this proposal, so as moderator I deem it to be accepted.

The CF minor version number should be incremented and Martin Juckes named as an additional author.

The second paragraph of section 3.5 of the CF standard document should be modified by including a new sentence after the existing sentence "The flag_meanings attribute is a string whose value is a blank separated list of descriptive words or phrases, one for each flag value." The new sentence is "Each word or phrase should consist of characters from the alphanumeric set and the following five: '_', '-', '.', '+', '@'."

The third point of section 3.5 of the conformance document should have the same new sentence added to it.

Jonathan