Jump to content

dvaosta

Community Members
  • Posts

    17
  • Joined

  • Last visited

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

dvaosta's Achievements

Discens

Discens (2/14)

3

Reputation

  1. This is the content of the user.cfg file (I omit the lobby credentials and the userreport.id): gui.splashscreen.enable = "true" gui.splashscreen.version = "1478828772" multiplayerhosting.port = "20595" multiplayerjoining.port = "20595" multiplayerserver = "127.0.0.1" playername.multiplayer = "dvaosta" playername.singleplayer = "DavideV" shadows = "false" showsky = "false" splashscreenversion = "0" userreport.enabledversion = "0" waterugly = "true"
  2. I never edited the config files. If you tell me where these files are located, I can see if there are strange values.
  3. You all have been joked by astronomers. Luckily, there is a group of illuminated people which know the truth: https://theflatearthsociety.org/home/
  4. Moreover, to join the path I propose to use the short range pathfinder (AKA vertex pathfinder), and CCmpPathfinder::ComputeShortPath takes as one of the parameters a maximum range, so if the unit is unable to reach the path with few "steps", it just give up and compute another long path. For example, suppose that you cache only paths longer than 50 tiles, and you allow to reuse a cached path only units that are max 3 tiles away from the starting point. You can call the short pathfinder with a range of 3-4 tiles, so if there are small obstructions the unit avoid them, but it doesn't walk for kilometers to join the path when there are large obstructions (walls, etc.).
  5. Yes. This is why I wrote that we have to check if the path can be reached from the current unit position. Obviously you can reuse a path only if the starting point is few tiles (2? 3? 4?) away from the unit and the target is far enough, otherwise the resulting path may be too different from the optimal one.
  6. For fast changing obstructions, the short range pathfinder is used (the long range one simply ignore other units). Regarding newly added buildings, I don't know if there are cases where the long range path is recomputed, but in the majority of cases also there the short range pathfinder is used to avoid them. I don't think that use something like a FIFO cache would be a problem. In fact, also when a unit move along a very long path, that path remain valid for a long time (and in the meanwhile several new obstructions can be placed).
  7. I've spent some time reading both this document and the code, and it seems that paths found by the long pathfinder are not cached but recomputed for each unit. Now, I think a cache can significantly improve performances, especially in the cases where you select a bunch of units that are close to each other and you order them to move to a point that's quite far: in these cases, the units tend to converge on a single line, so why to compute the path for each of them? So my proposal is to keep a cache of N paths (the most recent? The most frequently used? FIFO?); when a unit needs a path, we first search if in the cache there is already a path with the same target and a starting point that is less than X tiles away from the unit and we use the short range pathfinder to join that path (on which waypoint?), after having checked that that path can be reached, of course. Possible improvements: Reuse a cached path also if the end point in less than X tiles from the unit target (and there are not better cached paths); Reuse part of a path also if the unit position and target are near to intermediate waypoints of the cached path (here the difficult part is to efficiently search if a point is contained in a path); What do you think?
  8. A22 on win10. Samsung RC530 pc. Select the civic center (or another building where you can train units). Try to batch train units: it works. Press Fn+down arrow key: now the bath training doesn't work anymore. Press Fn+ left arrow key: it train 1 unit. Press Fn+up arrow key: sometimes, the civ center menu close, and if you click again on it the game minimize, but when you reopen it the bath training works again. Then there are some random variation: for example the batch stop working with Fn+up but works again after Fn+down, or it doesn't restart working anymore, etc.. Additionally, when batch training stop working, also action queueing stop working, so it seems that in reality the shift key is disabled.
  9. Version 22 on windows 10. Yes, I tried several times and it always happens.
  10. Launch 0ad, select tools and options -> options, click close; it will show the message box telling that there are changes and asking if you really want to close, also if you didn't change any option.
  11. Looking at the extinct volcano map, I noticed that, when the water rise, the ships can move across partially submerged buildings and trees (this is visible also in the trailer). My proposal is to implement 3D obstruction boxes and to compute which objects cross the water level each time the water rise (or, better, water level - x meters).
  12. Hi, in my opinion, I think that there are two things that are quite annoying when the player move around a map (especially if he's a beginner): 1. You can contemporary move by hitting a screen edge with the cursor or by using arrow keys: one of the first times I played, I was near the right border of the map and, using the arrows, I could move only along the border; it takes me some time to realize that the mouse cursor was on the right edge of the screen (so it also was not visible). I think that this can easily be prevented by disabling the movement with the cursor when using the arrow keys, and by highlighting the edge where the mouse cursor is (so that the player sees it and can move it before releasing the arrow key); 2. It's quite difficult to select one of the button on the top of the screen without hitting the border (and this happens especially when you are already busy and you don't want the map to move unexpectedly). This is probably more difficult to solve, but I think that wait a little (maybe making the delay configurable by the user) before moving if the cursor hit the top edge may be a good compromise.
  13. Thanks to both, now it works. Now I take a look to the code and maybe in future I'll be able to solve some of the simplest tickets.
  14. Hi, I downloaded the source code and I launched the update_workspaces.bat to create the VS 2013 workspace. I can correctly compile and launch it, and it works as expected. The problem is that in VS i cannot see all the files. For example, in atlas I see only the GameInterface folder and not the others (AtlasFrontends, AtlasObject and AtlasUI). I attach the VC++ Project file of the atlas project. My question is: is this expected, and I have to search these missing files in other projects or is it a bug? atlas.vcxproj
×
×
  • Create New...