I think I understand. IMO, it still feels worth it to me to maintain the old behavior, even if it had issues, and find a way to switch to new behavior and make the new behavior the default.
-Alex On 6/6/14 4:55 PM, "Michael A. Labriola" <labri...@digitalprimates.net> wrote: >>I haven't looked at the failing tests, but could it be true that those >>tests are not using XMLListCollection directly? They may wrap the XML >>in an XMLListCollection or just pass XML directly into the control where >>it gets wrapped, and then they manipulate the XML? I think there are >>lots of people doing that sort of thing so we should not break them. > >>You could be right that nobody really uses XMLListCollection today to >>add/remove items. If you want to gamble that that is the case, I'm >>willing to go along with that, but we should at minimum find a way that >>folks doing the XML manipulation directly don't get broken. > >Alex, > >What I was saying is that anyone who is using the XML directly would have >trouble also using the XMLListCollection with it in more than a basic >way. XMLListCollection actually changes the XML source in unpredictable >ways (re-parenting nodes, etc.) > >So, it's not that I don't think people are using it. I was just saying >that anyone who uses it in a more than a casual manner is either working >around those issues or (as we used to) is making a copy of the XML before >XMLListCollection is allowed to touch it so it doesn't screw things up >too badly. > >I don't know the right answer here. Honestly, I think the issue is that >the ListCollection views are trying to overlay a structure that doesn't >actually make sense onto XML. So, I would wager, people are mostly using >it to wrap XML so that it can be viewed in things like Lists/DataGrids. >They _may_ be adding and removing some nodes in a limited capacity, but >most likely they are playing around with the actual XML still since that >is the only way they could do e4x expressions etc. > >Mike >
