- Fixed: Artiflash played on initial save loading. I seem to recall this looking like an intentional change, but perhaps I broke it since it's completely pointless to play the animation only on the first load of a save game if nothing has been loaded beforehand.
- Reuse the old window in the SDL backend since in some instances switching windows causes issues (OS X fullscreen for instance (while using the SDL backend of course)).
- Clear the SDL render before copying in the framebuffer to remove HOM-like effect.
The new tag manager considers tag 0 'untagged' and won't create entries in its tag list for it, so the normal search algorithm can not find any such sector.
It now uses a linear search over all sectors instead, if tag 0 is looked for.
If the map name neither matched 'ExMy', 'MAPxy' or 'LEVELxy', 'checkstring' was left uninitialized before using as argument to 'strstr', leading to undefined results.
Spotted with Valgrind.
- Overrides the NOTELEPORT flag so actors with velocity don't need to disable it in order to perform A_Teleport, which could be screwed up by a teleporting line or sudden ACS interference.
- Ensure that the result is set to false if it fails prematurely. Wasn't sure if this was needed, but with the upcoming if/else statements, better safe than sorry.
- (int ptr = AAPTR_TARGET, state high, state low, float offsethigh = 0, float offsetlow = 0, bool includeHeight = true)
- Jumps if the pointer of the calling actor is higher or lower than itself, adding offsethigh or offsetlow depending on the circumstance.
- includeHeight works twofold.
- Includes the height of the calling actor if the pointer is higher to truly determine if they are completely above them or not.
- Includes the height of the pointer if the pointer is lower.
- Disable it to only check z differences without adding height.
Introduce AActor::TakeInventory, which unifies DoTakeInv from ACS and DoTakeInventory from Decorate, and AInventory::DepleteOrDestroy, which is extracted from the DoTakeInv core function, and use both where they're needed.
I don't know if the differences between DoTakeInv and DoTakeInventory were intentional, so I kept both behaviors.
The code never checked the starting position of the move and could be erroneously triggered in rare situations where the distance increased between actors but the hit boxes started overlapping because x or y distance got below the radius.
Changed it so that the code only gets executed when there's already an overlap before the move.
* added global functions that check whether FMod and OpenAL are present, without initializing the sound backend.
* make sound init code more fault tolerant. It will now try to switch between FMod and OpenAL if the currently active one cannot be found but the other one can.
* added 'ifoption' checks for sound backend to menu code.
* only show sound backends which are present and hide the options for the ones which are not.
This required the addition of a few exception handlers so to avoid #ifdef overuse I also added some #defines for non-Windows systems that allow using __try and __except directly in the code without #ifdef'ing them out.
UpdateSounds will not be called during screen wipes and the entire setup of this function suggests that this is not advisable at all.
The OpenAL stream updates were done deep inside this function implicitly.
This caused music to stop while a wipe was in progress. So in order to allow uninterrupted music playback during screen wipes the music updates need to be handled separately from sound updates and be called both in the main loop and the wipe loop.
I think that the OpenAL music updating should be offloaded to a separate thread but at least it's working now without causing interruptions during wipes.
- Viewheight change was delayed: Viewheight must be copied to player
structure as well as the PlayerPawn. Not sure if should actually use
deltaviewheight to spread the change out over a few tics instead of
being instant.
- Viewheight not preserved when travelling: player->viewheight must be
restored from pawn->ViewHeight, because the temporary player set it back
to the default viewheight.
- If palette index 255 happens to be white (e.g. as in Hexen), trying to
use white with DTA_FillColor would treat it as if you had never passed
it to DrawTexture().
- S_PrecacheLevel() must also mark currently playing sounds as
used. If we don't, the sound could be unloaded and the underlying
channel stopped without triggering a channel callback. That would leave
the code in s_sound.cpp thinking the sound is still playing even though
it isn't.
- Added an invalid channel check to FMODSoundRenderer::StopChannel() so
that orphan channels passed to it will be returned at least when
S_StopAllChannels() is called.
- OpenAL never actualy worked properly and was removed in later FMODs.
- Sound Manager was deprecated by Apple long ago and is not supported for
64-bit applications. It was also removed in later FMODs.
Tags are now handled by a tag manager class which stores sector/tag pairs. This way multiple entries can be added per sector.
Since UDMF does not have any arrays the additional tags are stored as a space separated string as 'MoreIDs'.
- When the OPL3 cores were added, DiskWriterIO was never updated to take
into account things like more than two OPL2 chips can be configured.
- DiskWriterIO no longer does any file writing directly. That function has
been split off into an OPLDump class, which has two specializations: one
for RDOS Play, and the other for DOSBox. If one chip is configured, it
dumps for a single OPL2, otherwise it dumps for an OPL3 (effectively
dual OPL2).
- TODO: Figure out why playback of raw OPL files doesn't sound nearly as
good as playing MIDI with the OPL emulation. It's probably something
simple I overlooked.
- Fixed: If you enlarged the game window (in windowed mode) so that the
window is bigger than the selected resolution, the menu would still take
its inputs from the portion in the upper left that matched the
resolution.
* make setting the line ID with P_TranslateLineDef explicit because there's one FraggleScript function that needs to work around the changes caused by this. There's also some functions setting only a temporary linedef. These would inevitably cause problems if the underlying data gets changed.
* remove FS function 'ChangeTag'. Fortunately this was just some long forgotten test stuff that can be removed without affecting any maps, but the feature would cause some serious problems in a more complex system.
With these changes it is guaranteed that after map setup the tag/ids won't change anymore.
This is done to encapsulate the gory details of tag search in one place so that the implementation of multiple tags per sector remains contained to a few isolated spots in the code.
This also moves the special 'tag == 0 -> activate backsector' handling into the iterator class.
