The description of the method stated that you could pass a string to look up an object layer based on its name, but the original code only looked for a gui property, regardless of type.
I'm using Tiled 0.11.0 on OSX, and my exported json-file doesn't contain a gui property on the object layers. Here's a snippet of my file:
```
{
"layers":[
{
"draworder":"topdown",
"height":0,
"name":"Objects",
"objects":[
{
"height":48,
"id":1,
"name":"pier",
"properties":
{
},
"rotation":0,
"type":"",
"visible":true,
"width":64,
"x":208,
"y":240
}],
"opacity":1,
"properties":
{
"type":"pier"
},
"type":"objectgroup",
"visible":true,
"width":0,
"x":0,
"y":0
}]
}]
}
```
The property in my file is called id and not uid. This might be due to changes in one of the later releases of Tiled. I kept the check for uid to avoid possibly breaking compatibility with older versions of Tiled.
I also added scaling of the sprite.
Feel free to make formatting changes, or rewrite the function to better fit the programming style of Phaser.
Since Tilemap.getTile lets you specify wether to return null
or an empty tile when index is -1, the WorldXY equivalent
should offer the same possibility for a more unified interface.
As implementation is leaning on 0.11.0 feature of Tiled, I made sure the earlier Tiled files that might be used are not going to break from this. Defaulting to empty string if the .type property is not defined by Tiled (for Tiled < 0.11.0), otherwise we use the provided value.
Why isn't the .type already read from the objects details as it's there? Latest release version of Tiled 0.11.0 has this field for Objects and it's quite extensively used by other frameworks.
Well, I need this and I suppose there are other people too who might be interested in having this in here too. Currently I go around this by creating a custom property of "type" and setting my value on it but that's just extremely silly.
- Memory optimization of delta-scroll shifting.
- The copy canvas is shared between all TilemapLayers
- The copy is done in segments to reduce the memory usage of the copy
canvas. Currently this is a 1/4 ratio.
- Device has the feature (by browser) check to see if bitblt works
- TilemapLayer will automatically enable the double-copy as needed
- Device.whenReady can be used to override the device value
- This change uses a secondary canvas and rectangle clearing instead of a
'copy' composition on the same canvas. This should hopefully address
the flickering issue in Safari and Safari Mobile
Addresses https://github.com/photonstorm/phaser/issues/1371 - where Tiled
used floor calculations and Phaser used round calculations.
There are no breaking changes with the demos; however, if there was code
that relied on behavior that _deviated_ from the defactor Tilemap behavior
then may be prone to "incorrect results" if using tileset images that are
not correctly aligned to tilesize multiples.
This also adds/corrects some warnings:
- A warning when the tileset size is not an even multiple (it was
suppressed due to using `round` prior to the checks.
- A warning if loading an image changes the cols/rows of a Tileset
- A specific warning for Tileset Image Collections which are not yet
supported
TilemapLayer
- Added "delta scroll rendering"
- Enabled (by default) via `renderSettings.enableScrollDelta`
- Dramatic CPU reduction: 30% to 2% scrolling 50x37 region
- Side-effect: fixed some Chrome de-opt issues with `render`
- Various rendering updates and removal of many `_mc` field
- Made `index` a read-only propert that reflects the layer.index
- Creates a local cache from tile index -> tileset, added `resetTilesetCache`
- Also prevents crash when an index not associated with a tileset is
used; see `missingImageOverfill`.
- Various debug options moved to `debugSettings` and simplified; documentation updated.
- Uses `debugSettings.missingImageOverfill` and `tileColor` (deprecated) to display invalid tileset images
- Only change the context alpha when the tile alpha changes; in Chrome 38
this removed a .5% penalty
- Added `renderSettings` which controls misc. rendering/optimizations.
- Updated documentation to @member, clarified
Tileset
- Added `containsTileIndex` method to allow being queried if the specific tile index is handled/supported.
- Uses a firstgid bias and interlacing to fit the data in single densely packed array.
- Updated documentation to @member, clarified
Discussion: http://www.html5gamedevs.com/topic/7409-tilemaps-changing/
It should remove the problem of updating stuff in a 100x100 tilemap making the engine recalculate every single update.(390 tiles changed in that map makes it a 100x100x390 loop.
Whould make the recalculate parameter in the setCollision functions unnecessary as well.
TilemapLayer.wrap property - if true the map is rendered as if it is on the surface of a toroid (donut) instead of a plane. This allows for games that seamlessly scroll from one edge to the opposite edge of the world without noticing the transition. Note that the World size must match the Map size (thanks @jotson #851)
Prior to this change, TilemapLayers always restricted themselves to
rendering strictly within the world bounds. If the camera was allowed
to go beyond the world bounds, then the tilemap would appear to stop
scrolling once the camera hit the world edge. This allows the tilemap
to continue scrolling, showing empty space beyond the edge of the map.
Additionally, when the new "wrap" parameter is true, the tilemap will
render its opposite edge in the empty space. This simulates the map as
if it was the surface of a toroid (donut) rather than a flat plane.
Member methods of Tilemap which take a layer parameter use getLayer on
that parameter to allow passing int, string, and TilemapLayer objects.
Tilemap.hasTile(...) should do so as well.
Tilemap.removeTileWorldXY lets you remove the tile at the given pixel value coordinates and updates the collision data.
If you pass `null` to Tilemap.putTile as the tile parameter it will pass the call over to Tilemap.removeTile.
