Writing yaml

General
1 
test(Id,Name,Description,IdFile,FileName,FileDescription,FileSize,FileType):-
    yaml_write(current_output,[thing{id:Id,name:Name, description:Description,data_distributions:[thing2{id:IdFile,name:FileName}]},thing{id:Id,name:Name, description:Description,data_distributions:[thing2{id:IdFile,name:FileName}]}],[]).

%%% ediprolog
%%%?- test(1,name,description,idFile,fileName,fileDescription,fileSize,fileType).
%@ - data_distributions:
%@   - id: idFile
%@     name: fileName
%@   description: description
%@   id: 1
%@   name: name
%@ - data_distributions:
%@   - id: idFile
%@     name: fileName
%@   description: description
%@   id: 1
%@   name: name
%@ true.

I expected:

%@ - id:1
%@   name: name
%@   description: description
%@   data_distributions:
%@   - id: idFile
%@     name: fileName
%@ - id:1
%@   name: name
%@   description: description
%@   data_distributions:
%@   - id: idFile
%@     name: fileName
%@ true.

My system expects the id to be the first item, then I do not think the order matters. Is there a way to get the yaml write to do this?

2

Your system is wrong, if it places significance on the key order.

E.g. Kubernetes will output YAML alphabetically sorted.

From a quick Googling:
https://yaml.org/spec/1.2.2/#mapping
“The content of a mapping node is an unordered set of key/value node pairs…”

3

Using yaml_write/3 to output a list, instead of a dict, would presumably retain the list order (I haven’t checked).

Here’s some code I wrote and have little-used, to convert between them:

% Recursively convert between Dict and pairs list
dict_pairs(Dict, Pairs) :-
    var(Pairs), !,
    % https://www.swi-prolog.org/pldoc/man?section=pairs
    dict_pairs(Dict, _, Pairs0),
    dict_to_pairs(Pairs0, Pairs).
dict_pairs(D, L) :-
    pairs_to_dict(L, D).

dict_to_pairs([], []).
dict_to_pairs([K-V|T], Pairs) :-
    is_list(V), !,
    dict_to_pairs(V, P),
    Pairs = [K-P|Pairs0],
    dict_to_pairs(T, Pairs0).
dict_to_pairs([K-V|T], Pairs) :-
    is_dict(V), !,
    dict_pairs(V, _, P0),
    dict_to_pairs(P0, P),
    Pairs = [K-P|Pairs0],
    dict_to_pairs(T, Pairs0).
dict_to_pairs([D|T], [P|Pairs]) :-
    is_dict(D), !,
    dict_pairs(D, _, P0),
    dict_to_pairs(P0, P),
    dict_to_pairs(T, Pairs).
dict_to_pairs([K-V|T], [K-V|Pairs0]) :-
    dict_to_pairs(T, Pairs0).

pairs_to_dict([], _{}).
pairs_to_dict([H|T], Dict) :-
    pairs_to_dict_([H|T], D),
    % Check whether list in K-V style
    (   D = [_-_|_] ->
        dict_pairs(Dict, _, D)
    ;   Dict = D
    ).

pairs_to_dict_([], []).
pairs_to_dict_([K-V|T], L) :-
    !,
    (   atomic(V) ->
        DH = V
    ;   pairs_to_dict(V, DH)
    ),
    L = [K-DH|Ds],
    pairs_to_dict_(T, Ds).
pairs_to_dict_([H|T], [D|Ds]) :-
    pairs_to_dict(H, D),
    pairs_to_dict_(T, Ds).
4

I actually think the software accepts the unordered list but as the yaml file is meant to also be read by people, and the reality I might have one hundred or more keys, it would good to retain the order, so the most pertinent keys such as id are at the top…

5

What’s stopping you from giving the humans a far more human-friendly output?

Do the humans need those 100+ keys detail?

6

It’s a good point.
But just trying to be consistent with how the files are used.
If there was a simple option I would use it, but if not then i’ll either live with it or write my own output format I guess.