Wow this week has flown by. It's seen a few delays as some unexpected issues cropped up, but overall things are going really well.

I know everyone's been waiting for the next release--especially those who've reported a couple of bugs that will be fixed in 508.1295. I think I can say with high confidence now that 508.1295 will be the first release of the DSified webclient. There are still a few more tests to do (isometric, for example), and a couple of minor things I'd like to attend to along with testing of more games. (Microlayers and contents order are still on my list, but I may release without them. I do however need to tweak the layer sort algorithm to take original order into account when layers are the same on the same object.) Culling out-of-bounds icons is important too, because it'll make things easier on the sorting algorithm. Also, I need some in-browser testing to be sure the Dart compiled to JavaScript still works. But I think now that I can truly say the day is close, and I'm aiming for early next week to get 508.1295 out--likely Tuesday.

Some of the hiccups I encountered this week were that "composite" images used for the status control and statpanels didn't work anymore, and a lot of mouse foo got broken--a problem I hadn't fully considered, since in the past all mouse commands basically sent screen coords to the server and said "Here, find me the icon-x/y params for that atom." In the new model that doesn't work, so there was a bit of scrambling.

The best news so far: The DSified webclient surpasses my expectations for server performance. Running Severed World, server-side CPU was extremely low most of the time--down to the 0-1 range for world.cpu, which for a 40 FPS game with a huge view is a huge win. I truly couldn't be happier at how well that's working out. The client-side performance could be a smidge better, which I attribute to the need to cull more icons, but that's something I'm sure I'll work out shortly.

Area appearances are working as expected. Atom and appearance garbage collection are in, including atoms used in the skin. (A new control function, atomsInUse(), can be used to get a list of all atoms the control uses. This returns null by default, since the skin will also look for any HTML elements with an "atom" attribute.)

Along the way I seem to have racked up a backlog of bugs to look at, which I still plan to do. I do have an actual list of things to still get to, so once the DSified webclient is in beta I can buckle down and investigate a number of those. I know there are a couple of things with the pager, and also I think it'd be good to look into possible ways to improve proc call overhead if possible--where even miniscule improvements could help a lot for games that make zillions of proc calls.

SS13 folks: One of the things I want to prioritize for 509 if I can is to improve resource delivery. That may be a can of worms, but it's been dying for a revamp, and I think it's time to start looking at it seriously to see if there's anything I can do to improve the process. There's already a thread on the Feature Requests forum for improving this, and I could use your help brainstorming. What I'd like to explore are ways of specifying resource delivery, offloading resources to mirror servers, and more. Some of this may also be of benefit to the webclient, as it might be possible to specify mirror locations for icons (the standard new-style .dmi files anyway), music and sound files, and more.