NewsForumsCodeLogsGallerySpecsSearch

List memory leak?

Community Forums/Monkey Talk/List memory leak?

sswift(Posted 2011) [#1]
Class List<T>

	Method Equals( lhs:Object,rhs:Object )
		Return lhs=rhs
	End
	
	Method Clear()
		_head=New Node<T>()
	End

        Method Count()
		Local n,node:=_head._succ
		While node<>_head
			node=node._succ
			n+=1
		Wend
		Return n
	End


First, why would you create a new node when you clear a list? If I clear a list, I want it to be empty, not have a single node in it with no data. Why not use null pointers to indicate an empty list like Blitzmax?

Second, if you simply create a new head node for the list when you go to clear it, where do all the old nodes go? They still reference eachother so the garbage collector should never collect them.

And lastly, that count code is weird. On first glance it appears it would count the head node and return an incorrect count but then I notice that it skips the head node, then skips the node after that, then begins its count. So it will arrive at the correct count, but it does so because it skips counting the first real node. I would have used Repeat Until, and put the node incrementor after the count increment, just in case someone decides to modify it later to change the nodes in some way.

Oh and what do lhs and rhs stand for?

Last edited 2011


Jesse(Posted 2011) [#2]


First, why would you create a new node when you clear a list? If I clear a list, I want it to be empty, not have a single node in it with no data. Why not use null pointers to indicate an empty list like Blitzmax?



I thought that was weak as well, he could have done it like this:
_head._succ = Null
_head._pred = Null




Second, if you simply create a new head node for the list when you go to clear it, where do all the old nodes go? They still reference eachother so the garbage collector should never collect them.



according to this post from Mark:

http://www.monkeycoder.co.nz/Community/post.php?topic=1269&post=11435
it's ok


And lastly, that count code is weird. On first glance it appears it would count the head node and return an incorrect count but then I notice that it skips the head node, then skips the node after that, then begins its count. So it will arrive at the correct count, but it does so because it skips counting the first real node. I would have used Repeat Until, and put the node incrementor after the count increment, just in case someone decides to modify it later to change the nodes in some way.



_head is used as the container for the nodes
_head._succ and _head._pred are the handles of the first and last node. It makes cense to me.


Oh and what do lhs and rhs stand for?


I think only mark knows.


skidracer(Posted 2011) [#3]

Oh and what do lhs and rhs stand for?



Left hand side, Right hand side.

Here are some notes from original Amiga Exec (kernal) manual that may show origin of pattern:

http://amigadev.elowar.com/read/ADCD_2.1/Libraries_Manual_guide/node02D8.html

Last edited 2011


therevills(Posted 2011) [#4]
LHS = Left Hand Side
RHS = Right Hand Side

Doh! Beaten by Skid

Last edited 2011