This just in via email:

So you’ve signed up to our super special Construct 3 newsletter, but we’re sure you’ve been thinking we’ve been a bit quiet. Well that’ll change soon, we promise. In the meantime, keep an eye on your inboxes and on our social media channels - some news will be coming your way on the 1st of February. Stay tuned.

Exciting times, can't wait!!

Crikey, better save some cash aside.

This is my crap game: https://play.google.com/store/apps/deta ... bbi.srduke

That was built 8 months ago or so with canvas+ and was/would be very slow with xdk.

no i havne't tried webview, it sounds like it sucks.. i have a pre android 5 phone so i know it will suck for my test device.. i'm ok with crosswalk and cocoon.io regardless of bloat.. i don't really care if a game is 30mb.. few people are going to not play my game because it's 22+ mb

there are way bigger problems to face than file size with mobile gaming

Is Crosswalk faster than webview?

yea just stick with what works and when it doesn't move on.. performance wise all i can say is design games that are not pushing cpu too hard.. consider ways of making games that don't need heavy gpu performance. I've made lots of games with cocoon and xdk (crosswalk) and they all run fine for everything I've done..

Have you tried webview by un ticking Cordova? I like the idea of not bundling chrome up with the apk, but the performance is awful in compared to canvas+, what are your findings?

I'd love to use xdk over cocoon, but I can't possibly see any reason to use xdk webview over canvas+ when the performance is awful, and this was with a bag-standard basic project. I used the .xdk file c2 made, is there anything else I should be doing to match canvas+'s out of the box performance?

Have you used arrays before? I just want to know how much detail to provide.

Yes, extensively, one of the few things I have somewhat of an understanding of

Okay, so let's say I've got these variables:

XPoint1

YPoint1

XPoint2

YPoint2

XPoint3

YPoint3

XPoint4

YPoint4

XPoint5

YPoint5

And I've got 5 behaviours that, currently, I have grouped into five different sets of events that, depending on the behavior, checks a system expression against it's relative number (behavior #1 checks for an overlapping point and stores it at XPoint1 and YPoint1, behavior #2 checks for an overlapping point and stores it at XPoint2 and YPoint2, etc). The five behaviors are familied. How do I efficiently, run a "for each "family"" call the relevant pair of variables that correspond to the behavior number in the family, or should I say, the loopnumber of the behavior?

So, if I have some variables that are named with different numbers (Height1, Height2, Height3, etc) how do I pick these with a loop index, for instance:

For each instance

Set variable Height'loopindex'

where the variables are each picked based on they're loopindex, because it now correlates to the same name?

rexrainbow

I've been tinkering with this effect and it's really cool! I'm having some trouble getting object positions to track accordingly. I'm using rex's behaviors but I'm using a tiled background for the object ground. Will that impact the way the perspective behavior works? I currently have grass (as the tiled background) and houses (as sprites) but I can't get my houses to appear anywhere after setting up the camera sprite to the ground behavior's camera and the houses perspective behavior to the grass's ground behavior parameters. In the sample capx it looks like 0,0 (on the X and Y parameters of the perspective behavior) is the upper left hand corner of the canvas but I don't see anything populating there when I test things out.

EDIT: I've also tried setting the logic positions to the sprites current x/y positions, which could be a mistake (as I'm not so confident with all the math), but I thought it was worth a shot. I also don't see them when I try that out.

It's been a while since I've use this but doesn't the math have to be inverted, or something?

The fact canvas+ (and really all non-browser engines) lacks so many features far outweighs any performance difference. Web Audio, form controls, XML parsing, standard-compliant inputs, sometimes even letterbox fullscreen, the list goes on and on. Fast and broken is worse than slower and working. If you can delete features to make something faster, it's not particularly useful if you need the features that got deleted. Note that (at least in the past I saw) it has other problems that come in to play as you increase the scale of your game, such as lacking the memory management that browsers have, so games tend to crash due to running out of memory where they would work in a real browser engine.

We deprecated non-browser engines a long time ago and I firmly believe no C2 users should use them, even if they're faster in some cases.

Modern webviews don't have any file size overhead. For example on Android if you target 5.0+, the APK is small compared to the size including Crosswalk. By the end of the year Android 5.0+ should be on the vast majority of devices. It's already at about 2/3rds worldwide.

Ashley I complied for webview on XDK, and for a simple platformer, the performance was far from usable, whereas canvas+ was perfect, speedwise. It completely screwed up alignment of tiled backgrounds and simulate key didn't work, but it was a usable speed, a solid 60 fps, rather than XDK's attempt. Is phonegap better for performance in your opinion?

awesome thanks for validating that! I remember trying that awhile back but yea it was HORRIBLE.. audio was f'd up and performance was terrible.. but that was way before android 5 was out

It still is horrible, XDK is no better in my opinion.

>

> *quote snip*

>

> I 100% agree with you, I also would have liked to have a third tilemap for jump-throughs, which would complicate issues even further, which leads me to suggest that tilemaps need the ability to set behaviors per tile/groups of tiles.

>

Well, that wouldn't really work, since adding behaviors to individual tiles would inevitably lead to people trying to add movement behaviors to them, and that wouldn't work for obvious reasons.

The best workaround I can suggest is to have 3 separate Tilemap objects, instead of 3 instances of the same Tilemap object. That way, you can filter the traces however you want.

I tried that, but I couldn't copy the tiles to the clones with JSON, but that's on a side note, I don't want to derail this thread.