We can use MergedStream to create a single combined stream with all the input and pass this to the hash function. This saves copying all the data into a MemoryStream to achieve the same goal, which requires more memory and allocations.
Callers can provide 3D points easily, and this avoids the need to allocate and slow down enumeration via the points.Select(p => new float3(p, 0)) wrapper.
Compiling these expressions is sadly expensive, and we needed new ones for every trait on every actor each time one was generated. The expressions thankfully can be shared as they are pure functions, which removes this overhead.
- Where it is accessible, use the length of the stream to presize the MemoryStream to the correct size.
- Instead of copying out the result via ToArray, grab the underlying buffer via GetBuffer and use that to create the sound source.
This avoids extraneous copying of the array containing the audio.
The render bounds for an actor now include the area covered
by bibs, shadows, and any other widgets. In many cases this
area is much larger than we really want to consider for
tooltips and mouse selection.
An optional Margin is added to Selectable to support cases
like infantry, where we want the mouse area of the actor
to be larger than the drawn selection box.
Avoid allocating the sheet builder each frame until it is needed. For mods that do not need to render models, this avoids allocating a large buffer and backing sheet as it will never render to the sheet. For mods that do render models, but don't need any this frame, this avoids allocating a new SheetBuilder that will not be used.
This prevents the capacity being set to 4 when the first item is added. For flat maps, the inverse projection will only ever be of size 1, thus this is sufficient capacity. For isometric maps, 1 is often sufficient, we only need more near height changes where the discontinuity means multiple cells may project back. We can pay for some reallocations to expand the size in these cases.
On flat maps, this reduces the memory required by the backing array 4x.
This avoids the allocations caused by LINQ when using traits.FirstOrDefault(Exts.IsTraitEnabled). This is important in FrozenActorLayer.RefreshState which is called very often. We apply the new helper method to all areas using the old pattern. An overload that takes an array allows arrays to be enumerated without causing allocations.
This allows callers to efficiently enumerate these returned collections without the allocation and overhead imposed by the IEnumerable interface. All implementations were already returning arrays, so this only required a signature change.
If the download failed, the map status would be MapStatus.DownloadError,
which would cause the Install method to return immediately. I've updated
the Install method now to account for MapStatus.DownloadError.
Additionally, internally renamed VisualBounds to SelectionOverlayBounds to avoid confusion with RenderBounds.
This step was necessary to prevent actors with selectable area smaller than their graphics to be removed too early from ScreenMap even though part of the graphics should still be visible.
RA cruisers were a prime example, but to a lesser extent several other actors were affected as well.
This separation also serves as preparation to determine the final RenderBounds from multiple source bounds later, to fix the remaining ScreenMap issues (building 'bibs', aircraft shadows).
- A 512x512 sheet is about half full after precaching and some usage, but uses 16x less memory than the default 2048x2048 sheet. This saving occurs twice as we maintain a managed buffer for this sheet.
- Only precache smaller fonts, as the larger fonts are less used and take up more space than is worthwhile.
- Only precache in white, as red is largely unused.
Packets from each chunk are now saved directly in an array, removing the overhead of a list. Additionally, a list is reused as a buffer for decoding packets into, preventing a new buffer from needing to be allocated for each chunk.
