Ashley's Forum Posts

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 19 Feb, 2014
  • Quote

  If you want to know what a feature does, look it up in the manual: https://www.scirra.com/manual/67/layouts

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 19 Feb, 2014
  • Quote

  STARTECHSTUDIOS - my best theory why you can't make a .capx showing collision cells are slower is because they're not slower! Alternatively maybe you're mistaking the latest version of Chrome's software rendered WebGL as a performance drop caused by collision cells, when it is unrelated.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 19 Feb, 2014
  • Quote

  If you press cancel with S_OK selected in test mode, it still succeeds. You need to select a failure return code.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 19 Feb, 2014
  • Quote

  It's just confusing UI in the test mode. If you press 'cancel' with S_OK selected as the return code, it still returns success because S_OK is selected. To simulate a failed purchase, change the status code to something like E_FAIL and press OK.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 19 Feb, 2014
  • Quote

  STARTECHSTUDIOS - I currently have no evidence (.capx examples) that collision cells reduce performance. I see no reason to create an option unless you show me this kind of evidence. We already added switches for the two behaviors which sometimes need it.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Does this really have much of an advantage over immediately unpacking parameters to local variables as the first thing 'On function' does? Both ways you have to create a mapping of index to name, just that this way you do that with conditions instead of actions. Local variables also allow for nicer naming in expressions (n instead of F.Param("n")) and are fast, whereas named parameters will require a lookup.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  It's probably just a mistake in your plugin. Nobody can help without more information about the error though.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Can't reproduce, works fine here so closing as probably a Chrome issue or something to do with your system. IE doesn't play anything because you don't have a .m4a version of the sound.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Closing, please follow the bug report guidelines.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Sorry, missed this thread originally: use the node-webkit plugin's 'show dev tools' action on start of layout and you can debug it like in Chrome. Note if there's a syntax error or something it won't get that far, but that should show up in the normal Chrome browser anyway.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Collision cells are a *huge* performance optimisation that make some games so much faster they go from unplayable to 60 FPS. It is well worth working around the minor edge cases where it can be slower.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Closing, please follow the guidelines and attach a .capx. Note IE does not support .ogg files and will only play .m4a if it is present.

• Try Construct 3

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

  Try Now Construct 3 users don't see these ads
• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  Well the parts about a HTML5 capable browser with latest graphics card drivers applies.

  Node-webkit is the same as Chrome, which is XP SP3+ I believe.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 18 Feb, 2014
  • Quote

  No, it's just that old mobile hardware is heavily fillrate limited. 3D games are rendered differently to 2D games and are actually lighter on the fillrate. Due to depth testing, 3D games can more or less write to every pixel on screen just once. However 2D games with a back-to-front renderer like Construct 2 has with lots of overlapping objects hits the fillrate hard, and old GPUs can't push that much bandwidth. Newer devices have no trouble though.

• • Ashley Construct Team Founder
  • 
    Moderator
  • • Joined 21 May, 2007
    • 
    • 672 topics • 31,115 posts
  • 1

  • 17 Feb, 2014
  • Quote

  pixel perfick - the performance you describe strongly suggests software rendering. Perhaps you have old drivers which are blacklisted, did you try updating them? Also did you try Firefox and IE and were the results consistent across them? If you only tested Chrome, note the current version has an issue where it may use software-rendered WebGL which is horribly slow. The next Chrome out in a few weeks should fix that.

  Juryiel - the 'high DPI' mode does not alter the CPU load, only the GPU load. I think this is strong evidence the game is bottlenecked on the GPU hardware capabilities, and therefore a native engine would not be faster.