This was done to ensure that this code only runs when the thinker itself is fully set up.
With a constructor there is no control about such things, if some common initialization needs to be done it has to be in the base constructor, but that makes the entire approach chosen here to ensure proper linking into the thinker chains impossible.
ZDoom originally did it that way, which resulted in a very inflexible system and required some awful hacks to let the serializer work with it - the corresponding bSerialOverride flag is now gone.
The only thinker class still having a constructor is DFraggleThinker, because it contains non-serializable data that needs to be initialized in a piece of code that always runs, regardless of whether the object is created explicitly or from a savegame.
This doesn't really write out any info for the pointer, if the level does not match it just errors out.
This is both for quick detection of badly used level data and for automatic restoring of the pointer from the serializer's working level.
This also removed the temporary workarounds in DAutomap and DLevelScript to restore these pointers when a savegame is loaded.
Since the SpawnedThings array is still available when polyobjects are spawned it makes no sense to create an expensive linked list in P_SpawnMapThing.
This can be done far better by scanning through the array again and collect all matching items in a second array.
This setup has been a constant source of problems so now I reviewed all uses of FName to make sure that everything that needs to be initialized is done manually.
This also merges the player_t constructor into the class definition as default values.
This was done mainly to reduce the amount of occurences of the word FTexture but it immediately helped detect two small and mostly harmless bugs that were found due to the stricter type checks.
- added a few access functions for FActorInfo variables.
With PClassActor now empty the class descriptors can finally be converted back to static data outside the class hierarchy, like they were before the scripting merge, and untangle the game data from VM internals.
- disabled the Build map loader after finding out that it has been completely broken and nonfunctional for a long time. Since this has no real value it will probably removed entirely in an upcoming commit.
Daedalus triggers this with a 0x85 character which in Windows CP 1252 is the ellipsis (...) The converter will assume ISO-8859-1, though, but cannot do anything with these characters because they map to the font being used here.
* everything related to scripting is now placed in a subdirectory 'scripting', which itself is separated into DECORATE, ZSCRIPT, the VM and code generation.
* a few items have been moved to different headers so that the DECORATE parser definitions can mostly be kept local. The only exception at the moment is the flags interface on which 3 source files depend.
I wish I had realized this the last time it came up - it would have saved me a lot of trouble.
But as it turns out, the more recent travelling code makes all of this completely unnecessary, working perfectly fine with deleting the player pawns along with the rest of the thinkers before loading the stored ones from the savegame (and getting rid of those in G_FinishTravel.)
And with a sane savegame format that does not depend on side effects from how the thinker serializing handled linking into the lists the old code was even harmful, leaving voodoo dolls behind.
I had the exact same effect when I tried to reshuffle some things for reliably restoring portals, but did not make the connection to interference between two mutually incompatible player travelling mechanisms that just worked by sheer happenstance with the original order of things.
- changed S_GetMusic to return a const pointer to the actual music name instead of a copy. The only thing this is used for is the savegame code and it has no use for a copy, it can work far more efficiently with a const pointer.
After testing with a savegame on ZDCMP2 which is probably the largest map in existence, timing both methods resulted in a speed difference of less than 40 ms (70 vs 110 ms for reading all sectory, linedefs, sidedefs and objects).
This compares to an overall restoration time, including reloading the level, precaching all textures and setting everything up, of approx. 1.2 s, meaning an increase of 3% of the entire reloading time.
That's simply not worth all the negative side effects that may happen with a method that highly depends on proper code construction.
On the other hand, using random access means that a savegame version change is only needed now when the semantics of a field change, but not if some get added or deleted.
- do not I_Error out in the serializer unless caused by a programming error.
It is better to let the serializer finish, collect all the errors and I_Error out when the game is known to be in a stable enough state to allow unwinding.
It turned out this may not be done automatically when opening the savegame - it has to be done later, after the pre-spawned map thinkers and all connected objects have been destroyed.
The object deserializer also has to be rather careful about dealing with parse errors, because if something goes wrong a whole batch of uninitialized or partially initialized objects will be left behind to destroy.
This means that no object class may assume that anything but the default constructor has been run on it and needs to check any variable it may reference.
