Ashley's Forum Posts

  • The problem started with Chrome 38 so based on their 6-week cycle I'd guess it will be fixed somewhere around Chrome 40-42 in a few months, but it depends on if they've fixed it already.

    The best thing to do is to star the bug report and voice your concern in the comments. The main bug report for the issue appears to be this one: https://code.google.com/p/chromium/issues/detail?id=422000 (Google are closing other reports as duplicates of this one)

  • As far as I can tell this is the ordinary IE browser engine (Trident) given a new UI. So it's just a new skin on the same old engine.

    However I think the real interesting opportunity is for this new browser to move to a rapid release cycle. That would be great news for ensuring IE progresses as fast as other browsers. However I've also read the bog standard IE11 will still be shipped with Windows 10 for backwards compatibility, so we'll still be juggling various IE versions for some time to come...

  • You do not have permission to view this post

  • IE doesn't support downloading local data. If IE gets support for the <a download> attribute, it will work.

  • I think a better solution would be improved tools to find all the references to an object, and then the editor can show you what's about to be removed before you remove it. (It's also undoable so if it suddenly breaks your game you can put it back.)

    Allowing events for deleted objects is pretty complicated, since if you try to edit an event for a deleted object then I guess you want to still be able to see all its behaviors, families, container, variables etc. Also if you do actually want all its events removed, then you have another boring manual step of sifting through the project deleting all the missing-object events.

  • shinkan - as far as I can tell from your example, your process is to:

    1) load 100mb of images on startup

    2) draw them in to Paster (which only uses ~16mb)

    3) (the request) release the 100mb of images

    Even if this was supported, it appears your peak memory usage is still 116mb. If the device is memory constrained, then unloading the textures afterwards will not help, it will have already risked running out of memory at the point of drawing them in to Paster. In fact using Paster increases the peak memory usage from 100 to 116mb. And if it can handle having 100mb of images loaded for a moment, then surely it can handle leaving them there until the end of the layout too? Since AFAIK the main issues with memory usage come from crashes due to peaking high, it seems the better option is to give up the optimisation and go with a lower steady 100mb than a 116mb peak, plus it is probably simpler.

    I don't think everyone else who's agreed so far is aware of this. I think there needs to be a description of a better use case where objects are created/destroyed during runtime and they really can't hold their images in memory until the end of the layout.

  • Closing, there is nothing we can do about reports which don't include steps to reproduce, especially if it takes hours to reproduce (debugging an issue often involves trying it 10-15 times with changes each time so this would take weeks and weeks by the sound of it).

  • It doesn't matter. Construct 2 automatically generates spritesheets on export.

  • Closing, please follow the bug report guidelines or there is simply nothing we can do about this.

  • Closing, please follow the bug report guidelines and attach a .capx, or there is basically nothing we can do to investigate the problem.

    Please also try disabling any browser addons in case they are breaking it.

  • Aphrodite - it's just a quirk of the test, you can ignore it.

    TiAm - on a high end phone it should have no problems at all with the fillrate.

  • I haven't noticed any issues at all. sbperftest gets a great score on my new Moto X (2nd gen) which sounds like it has the same hardware as the G3.

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads

  • r193 will probably be out early next year.

  • You need to be able to identify which system setups are affected and submit bugs to Google with minimal demos that reproduce the problem. The Chrome GPU team are always making tweaks to try to make the best tradeoff between rendering and input latency, and I've seen they've just introduced a new RendererScheduler to Canary which may affect this, so it seems to be a work in progress still. However giving real-world examples to the Chrome team should give them something to test their changes against.

    I think the best way to measure input latency is with a stopwatch it's hard to automate this kind of thing since it's a long chain through the system and code will only see it's stage in the chain.

    Firefox Nightly is way improved for me here so that may be worth looking at.

  • 12mb isn't very big even for mobile, but if it bothers you, have a read of tips to reduce the download size.