 |
lua-l archive |
|
|
||
|
It could be considered radical maybe for tables.for userdata and lightuserdata, it makes perfect sense to use the same syntax to copy one userdata to another one the same as you would a string value, rather than a function to do it.
keep in mind this would be controlled by the existence or not of the __assign metamethod. Presumably whoever would use this would do so judiciously.
Looks like I don't actually need __assign (yet) for my application, as all the things i am assigning belong to a parent table, so I get the same effect with a __newindex on the parent.
but it doesn't really make that much sense that lua exposes all the other operators except assignment.
I've already coded part of the __cast metamethod. It's working very well, I'm able to automatically convert tables into strings with a callback now.
Actually I'd be happy to forgoe all this if the table concept was extended instead to have a single unindexed value which would be used if there was no index made.
e.g. a la windows registry. You can set the value of a key, as well as members in a key. It's the unnamed index/default value.
Adrien Aaron Brown wrote:
Adrien de Croy wrote:this would also then allow lua script like function dosomething(object1, object2) object2.somemember = "hello"; object1 = object2; -- deep copy in host via __assign -- metamethod rather than table reference endI think this is a bad idea, because it's a radical change to Lua's semantics, in which an assignment whose left-hand side is a variable name always mutates the variable, never the value contained in it. An __assign metamethod would change that and also block regular assignment. Better to do something like: deepcopy(src, dest) -- Mutates dest into a deep copy of src. -- Aaron http://arundelo.com/ _________________________________________________________________A new home for Mom, no cleanup required. All starts here. http://www.reallivemoms.com?ocid=TXT_TAGHM&loc=us
-- Adrien de Croy - WinGate Proxy Server - http://www.wingate.com