I've found, on mobile, rotation is something to be very aware of. As in, on a lower end device, even a handful of rotating objects can incur a massive performance hit, especially if you are rotating them every tick.

This isn't some shortcoming of C2 either: rotation is just a very expensive operation to perform compared to, say, moving an object around.

There's plenty of other good advice for mobile, but this is something I haven't seen mentioned very often, even though it can grind a game to a halt.

Perfect timing for asking delaflaquita Intel integrated the expansion file plugin into the XDK build system today!

Tomorrow I will run few tests and report.

Wow, really? This would be awesome for the game I'm working on now, which includes some vids. Really had to squeeze to get it in under the 50mb limit. I'd love to be able to up that 10mb or so; would make a world of difference to the video quality of the included clips.

So...how do we go about using this?

I used the tilemap object. I've never needed any particularly sophisticated tools, so maybe my case isn't a good example. In the game I'm thinking of, IIRC, I had auto-tiling for a couple tiles types (not gonna lie, that was difficult since I hadn't dealt with bitwise logic previously), plus copy/cut/paste function, and horz/vert flip. That's about it.

Unfortunately, that code is not very modular (early project), so I can't exactly strip it out easily for a demo capx. Also, I never worked on a way to migrate levels back into the editor, which, I can imagine, would take some doing.

EDIT: Okay, so here's an idea: A "Load Tilemap" option in the Tilemap object Properties. It let's us specify a project file that contains JSON representing a C2 tilemap. Then we can create tilemaps with an in-game editor, export as JSON, and bring the new layout back into C2 with minimal fuss. Couldn't this be done within the scope of C2?

For mobile, spritefonts for sure. Faster and more reliable.

Full Box2d Web physics, instead of the current physics setup which is limited like a demo version.

Spritefont+ made native.

A Tween behavior like LiteTween.

+1000

I'd love to see improvement to tilemap, but I don't think it's realistic to expect any UI revisions til C3. In my case, I don't use tiles as often as some, but when I do, I make my own 'in-game' editor depending on my needs.

On the other hand, the addition of Litetween/Spritefont+/Paster would be enormously helpful, and wouldn't require any monkeying with the UI.

Litetween would probably top my list, because it is just so crazy useful, and would be easy to maintain since all it does is manipulate values.

I don't think these need to wait til C3...especially if it's going to be 1-2 years before we get our hands on it.

Wow, I'm surprised he even took that seriously, the email looks like spam or some kind of prank (though, I guess it wasn't).

Anyway...yeah, I guess he infringed, but going after a guy like that is like a police officer screaming in the face of a little old lady because she double parked.

What a bunch of pond scum.

+1

I almost never use pin for this very reason. Easy to 'pin' with event logic, and it works the way I want (same tick).

Wow, why didn't I come across OBS when I was trying to find a desktop recorder a few months ago? Open source FTW!

FYI, doesn't sound like the '10,000 limit' affects anything at all. C2 doesn't generate icons for every frame of every animation, just for the first frame of the default animation. You are only going to hit a limit if you have 1000's of different object types...which truly would be crazy.

—

It sounds like you've basically answered your own question: since C2 spritesheets on export, you are never realistically going to hit the chrome 'limit' of ~6400 images. But, you are hitting it on preview since each frame is being treated as a separate image.

Ashley

The solution to this would simply be a project properties option (default=FALSE) that would force the editor to spirtesheet objects prior to preview. The icon limit is irrelevant: nobody would hit it without thousands of objects. Problem solved.

Ashley

Would it be possible to get a copy of the sbperftest capx?

I ask because I would like to compare performance/stability across crosswalk versions, namely 7, 8, 9, and 10. This is related to the infamous janking issue, on which we finally seem to be getting some traction. I'm trying to test a theory that this originated with Chromium v38, and that the regression spanned multiple platforms.

If, for whatever reason, you would prefer to keep that test internal, I understand. I can mock up something on my own, but this would sure save time.

cesisco

Maybe this could have been handled better, maybe not. But I certainly don't hold it against Ashley.

When this bug first occurred, he filed his own report (which collected dust), and was active on the report this thread links to (which was also largely propelled by someone called Jerry Jongerius, who has probably done more legwork on profiling and tracking this bug than anyone, and who, as far as I know, has no affiliation with C2 or Scirra). Ashley also created and refined the sbperftest for evaluating jank, framedrops, etc., with a specialized demo run of space blaster. Oh, and of course, he also develops C2 and somehow finds time to be active on the forums. He may even sleep.

I, on the other hand, spent 5 minutes writing a thread, which just happened to take off. BTW, neither of those twitter posts are mine; one is from Scirra (Ashley) itself. The thanks is largely due to the C2 community, which, as always, rose to the occasion.

I think at the beginning, we all underestimated this bug, and sort of assumed that it's severity would ensure it would be fixed quickly. It wasn't, and while the current stable is much improved, we're still trying to stave off future regressions.

The main goal in mind for me is to indicate to google that display quality is not a trivial issue for Web 3.0 (Browser-as-OS). There needs to be better internal testing to ensure that regressions like this don't make it into stable versions in the future...and that they are fixed quickly if they do, as with security issues and other major gaffes.

The best thing for this is competition. Ironically, the next contender may be everyone's old frenemy Microsoft. Reading between the lines of their recent announcements, I think it's very likely we may see something similar to node-webkit/crosswalk built around Spartan, where we can package apps with a high-performance browser runtime seamlessly and across multiple (MS) platforms. On windows at least, this would be a welcome addition.

cesisco

Thanks! That was exactly what I was looking for.