Eli Zaretskii <eliz@gnu.org> writes:

> Thanks, please see a few comments below.

Thanks for reviewing.

> json-serialize-use-plists is the literal name of a variable, so it
> should have the @code markup, not @var.
>
Thanks for these doc comments, but I've gotten rid of
json-serialize-use-plists, after coming to the conclusion that the map
type (hashtable, alist, or plist) can be reliably auto-detected in
json-serialize.  As a consequence, less stuff to document..


>> +          if ( EQ (Vjson_serialize_use_plists, Qt) ) {
>> +            key_symbol = XCAR (tail);
>> +            tail = XCDR(tail);
>> +            CHECK_CONS (tail);
>> +            value = XCAR (tail);
>> +            if (EQ (tail, li.tortoise)) circular_list (lisp);
>> +          } else {
>> +            Lisp_Object pair = XCAR (tail);
>
> Our style is to write
>
> [...]
>
> Also, please don't leav any whitespace between the opening parenthesis
> and the following character, and similarly with closing parens.  IOW,
> this:

OK, got it.

> Here I'd use intern_1 instead, it would allow you to avoid
> unnecessarily consing Lisp objects.  (Yes, I realize that the same
> comment applies to the existing code.)

Riight, intern_1 sounds good... I know I can't just malloc() stuff as
usually right? But also, I have no idea what I'm doing, I aped this from
somewhere where it looked more-or-less responsibly done.

  json_object_foreach (json, key_str, value)
  {
    USE_SAFE_ALLOCA;
    char *keyword_key_str = SAFE_ALLOCA (1 + strlen(key_str) + 1);
    keyword_key_str[0]=':';
    strcpy(&keyword_key_str[1],key_str);
    Lisp_Object key = intern_1(keyword_key_str, strlen(key_str)+1);
    result = Fcons (key, result); /* build the plist as
                                     value-key since
                                     we're going to
                                     reverse it in the
                                     end.*/
    result = Fcons (json_to_lisp (value, object_type), result);
    SAFE_FREE ();
  }

New patch after my sig
João