Sébastien Cevey wrote:
Hello folks!
Hello!
[here was a nice high level motivation description]
Here follows some quick comments.
Here is my proposed API, strongly based on anders' ideas:
/* Prototypes */
xmms_value_list_t * xmms_value_list_get_iterator (xmms_value_t *val);
int xmms_value_list_get (xmms_value_list_t *it, xmms_value_t **val);
int xmms_value_list_valid (xmms_value_list_t *it);
void xmms_value_list_first (xmms_value_list_t *it);
void xmms_value_list_next (xmms_value_list_t *it);
int xmms_value_list_position (xmms_value_list_t *it);
I think I would s/xmms_value_list/xmms_value_iter/
Is xmms_value_list_position useful?
[]
Notes:
- There is no reason why xmms_value_list_get() should return false,
except if you pass it a crap 'it' argument. I think.
When outside list?
(which would have the side effect that you could do:
while (xmms_value_iter_get (it, &v)) {
do_stuff (v);
xmms_value_iter_next (it);
}
)
- We can have recursive lists/dicts/etc with that.
\o/
- We could probably add a foreach function, as well as convenience
aliases (_goto_index (it, pos), _last (it),
xmms_value_list_get_int (it, &i), etc)
I think _last should be part of the low-level api. (i.e internally lists
have pointers to last argument aswell)
- position here is in the interval [0; list_length]. if
position==list_length, the iterator is one step beyond the last
element (you cannot advance it more).
As I said above I'm not sure the function for getting iterator's current
position is useful. As for the convenience function _goto_index yes I
agree (or maybe not on the ability to position it outside the list)
That was it for accessing the list!
Or was it? :)
You are forgetting (intentionally or not) the ownership/referencecounting.
Does the "list" value own the iterator? Are values you get from
iter_get() borrowed or do you need to unref them?
Now on to the API to modify to the
list!
void xmms_value_set_empty_list (xmms_value_t *val);
I guess this means that we have similar api for all types:
v = xmms_value_new ();
xmms_value_set_int(v, 1337);
And not:
v = xmms_value_new_int (1337);
That means that we can initialize a "typeless" value_t? Do we really
want that.
Having no "xmms_value_set_int" means that the _same_ value_t can be used
for all ints with same value.
Lists are of course a different beast as they have to be mutable. But
keeping the api same and thus having a xmms_value_new_list() is better imho.
Unless you have some arguments where splitting allocation of value_t and
setting its value is useful.
[]
- _insert inserts the given value at the current position, thus
shifting the current item to the right. it also works if the
iterator points to the end of the list: you shift the end and insert
the value instead (AKA append).
- We can have various aliases for lazy people, as shown above.
- Do we need any more basic functions than those two?
I think those are enough.
This is only for lists; dicts and propdicts will each have their own
iterator type and functions, which I will detail later.
Or will they?
Maybe dicts and propdicts just are lists of pairs?
Then just some convenience functions to find stuff in the list.
anders
--
_______________________________________________
Xmms2-devel mailing list
Xmms2-devel@lists.xmms.se
http://lists.xmms.se/cgi-bin/mailman/listinfo/xmms2-devel
