- From: Claude Zervas <claude@utlco.com>
- Date: Fri, 13 Nov 1998 15:15:41 -0800
- To: www-dom@w3.org
At 05:21 PM 11/13/98 +0000, Miles Sabin wrote: >John Cowan wrote, >> Miles Sabin wrote: >> > Even with a remove() operation on the iterator class, this >> > will still cause serious problems, because i will be >> > invalidated every time a node is removed via j. >> >> Agreed. But other solutions are hardly better, since they >> involve using remove operations with at best O(log n) behavior. >> >> 1) You can't win; >> >> 2) You can't even break even. > >Yes you can ... > >We just have ultra simple iterators, and *document* the >conditions under which they remain valid, and rely on >programmers being smart enough to know when those >conditions hold and when they don't, or being able to >ensure they hold (with manual updates or synchronization, >or whatever). > I completely agree with Miles on this. As long as any gotchas are well documented, it shouldn't be a problem. People can always do index-based traversals using live NodeLists if iterators seem too scary. Regards, -Claude Zervas
Received on Friday, 13 November 1998 18:25:05 UTC