The core problem I fixed was in CardView's (missing) destructor - on construction, a CardView will set itself as a member of its parent CardInstance, so it stands to reason that when it's about to be destroyed, it should do the inverse and remove itself in the same fashion from its parent. This explains why weird graphic glitches were seen when casting Animate Dead on cards in a graveyard - the position data it was trying to use was already deleted from memory (a cardview is deleted on cleanup at the end of a turn if it's gone to the graveyard), but no one nulled out the deleted card view reference from the instance, so we'd access invalid data.
Some peripheral changes in this checkin: two helper functions in CardGui (GetCenterX, GetCenterY) that are part of my navigation patch are included. They're unused in the current code base, so this has zero impact. (I'm only checking them in as it's more work than it's worth to refactor them into a separate changelist. The core of the nav patch requires my mods to Closest.cpp / CardSelector.cpp to have any effect.) I also included a helper function in the debug routines to spit out hex pointer addresses in trace outputs, which I used to chase down this bug.
I've patched the problem so that ActionLayer checks for a valid ID before doing anything, and does a no-op for IDs it doesn't know how to handle. However, this only fixes the problem during gameplay. It's quite possible that hitting the triangle button in other places in the app might equally cause a crash.
Issue: 544
While debugging this, I noticed a separate issue: when changing profiles, we'd actually call refresh twice. Removed the spurious call, as reloading profiles doesn't need to concern itself with the image cache - that's already covered by the game options menu.
Also did some minor formatting / cleanup in the JGfx code for PSP - stubbed out a bunch of JLOG calls I had put in while debugging the PNG loading code.
This is the main fix for the situation where we try to load PNGs after the image cache has been saturated - instead of requiring 1 Meg as a temporary buffer to perform the swizzle operation on a full image (ie 512 * 512 texture size), these mods allow the swizzle to work in line with a temp buffer of 512 * 8 lines (ie 16K of memory). I also fixed a case where the swizzle didn't operate correctly on PNGs that weren't an even multiplier height of 8 - we'd drop the remaining lines on the floor, creating garbage lines at the bottom of those images.
1) Consolidated the Loggers so that we only have one in JGE. Added a helper constructor/destructor to the JLogger class so that you can instantiate one at the top of a function, it'll trace out a 'start' and 'end' message when it goes in & out of scope.
2) Fixed the crash part of a bug I've been chasing down where, on the psp, after saturating the cache, the game dies when trying to reload the background PNG image of the deck menu. We still need to fix the root cause of the failure (not enough memory to allocate a temporary buffer for the swizzle operation), but at least the psp doesn't lock up anymore. I've also left behind all the log traces I inserted into the LoadPNG code, since we'll probably need them again.
- fixed the debug framerate calculation / drawing code (consolidated duplicated functions & variables mDelta vs a win-only mDeltaTime, no reason to have two), repositioned it to draw on the bottom left (it was colliding with other debug text for the cache info).
- for debug testing purposes, added a #define override to force the image cache to work with less memory (I've picked 8 megs, which is what the psp seems to use - the normal default on win/linux is 20 megs).
Two things: 1) there was an extremely inefficient implementation in the particle system of a queue where, when particles expired, each expired element would be copied over(replaced) by one at the rear of the container. Depending on the performance of the update, this could range anywhere from 100 to 500 (the max particle count) memcpy operations per update loop. I replaced the flat array with a std::list, and simply pop the unneeded elements when they expire now. This seems to shave ~30% off the time spent in the Update() call.
2) Hardcoded the number of emitters for the mana particles to 60 (the default in the psi file seems to be around 114). This reduces the amount of iterations per tapped mana by just over half, which also helps the performance, and it's not really noticeable - the glow effect is a *tiny* bit more muted, but chances are, if you didn't read this comment, you probably wouldn't have noticed.
I've gone from having my psp start lagging noticeably at 8 tapped mana to about 12 - at this point, I think the bigger remaining lag is the fact that when there's a cache miss for a card image, we're doing synchronous i/o on the main thread - the framerate drop this causes is far more noticable than the mana particle lag.
- Added a possibility to put a file "Res.txt" instead of the folder "Res". The file Res.txt is a simple 1 line text file, telling where to find the Res folder, terminated by "/". For example: "../../wagic_res".
This addresses issue 428 . This could also help us in the future, to develop mods.
I'm also noticing that the GetQuad(int) variant never seems to get hit, so I suspect that the ID lookup map is redundant. I left it alone as the JResourceManager base class forces the need for the function; I need to spend more time looking at just how much of JResourceManager we actually use at this point.
Also renamed WindowRect to windowRect & PixelFormat to pixelFormat for consistency with other locally declared variables. (Uppercase should be reserved for class names, ie WindowRect is the class/struct, windowRect is an local object in your code.)
