* Optimisations in a bunch of places, including to static
* Fix some visual errors with text and patches near the edges of the screen in non-standard resolutions.
* Changed the save format a little more, since I had free reign over it.
* Modified the string drawing functions to not awkwardly clip at the very edges of the screen, considering the relevant patch drawing functions avoid out-of-range memory writes.
* Fix a bug where I got emeralds and mapnumbers mixed up when checking for the "there's a tailsbot" flag.
* Consider a save with an invalid skin an invalid save, rather than allowing its continued presence.
* Each time you die, the number of game overs you've had is counted.
* Your save file updates to record this.
* The number of startinglives is determined by the number of times you'ved game-overed, with the maximum being infinity lives (thereby providing a cap on the number of game overs you can go through in a typical game).
Requires a new patch.dta, but I'm not uploading that yet because not happy with the icon we've got going for infinity lives on the save select menu.
Lua lump load only
This branch prevents the following Lua functions from being used within hooks or coroutines, to prevent netgame desync problems or any other similar oddities when using these weirdly:
* `freeslot()`
* `addHook()`
* `hud.add()`
* `COM_AddCommand()`
* `CV_RegisterVar()`
All of the above should only be called when the Lua script is being loaded (i.e. it takes place the same instant a ".lua" file or a WAD file with a "LUA_" lump is being loaded)
To put it another way:
```
//This is valid usage:
freeslot("MT_TEST")
//This is NOT valid usage
addHook("MobjSpawn", function() freeslot("MT_TEST2") end, MT_PLAYER)
```
See *Monster Iestyn/Lua lump load only* on the FTP for an exe and test script.
See merge request !87
Add-ons menu
Does what it says on the tin! Very pretty. Would like help testing in filesystems that are not Windows, although I've recieved assurances from Sryder that since it's based on the same filesrch code, it should work cross-platform. Just want to confirm it works, though.
Also, M_* is whitelisted lumpname too.
root/!LatestSRB2Files/srb2win_branch_addfile.exe, requires new patch.dta.
See merge request !92
Consistent flat alignment
Does what it says on the tin. Consistent between the three different plane drawers:
* Software flat (previously the only one working as intended)
* Software sloped (took a lot of work)
* OpenGL flat and sloped (worked reasonably well but used different signs for some reason)
Check out root/!LatestSRB2Files/srb2win_branch_flat2.exe, root/toaster/flatalignment.wad and any in-dev DSZ1 to test it all out.
See merge request !78
NiGHTS Paraloop-special emblems
Implemented the two NiGHTS Emblem types in the TUDD via var modifications to ET_GLOBAL!
* GE_NIGHTSPULL (for it to chase you once paralooped) - Sun emblem
* GE_NIGHTSITEM (for hidden until paralooped) - Moon emblem
* Combined means that you essentially have to loop it twice.
Also, revamped NIGHTSITEM to suck a little less (and have more compatibility with MT_EMBLEM).
Check out root/!LatestSRB2Files/srb2win_branch_nightsemblems.exe with the latest patch.dta, along with root/toaster/nightstest.soc for a sample in GFZS.
See merge request !94
MF2_LINKDRAW reliability improvements
Seriously improves the reliability of the linkdraw system (in Software, the only rendering engine which actually does anything with this information), reducing the number of visual errors it produces significantly. Deliberately engineered to only affect objects with MF2_LINKDRAW applied.
Of particular note: Standing in shallow pools of water as Smiles will now reliably associate the bottom half of the tails following mobj with the underwater portion of Smiles' body, and the flicking ends with his un-submerged head.
Test with ```<root>/!LatestSRB2Files/srb2win_branch_linkdraw.exe``` and ```<root>/toaster/smiles.wad```.
See merge request !101
Wallspike tweaks
Sorry, MI. You told me to test the objects via objectplace, and they worked fine there - but @FuriousFox had some [issues](http://i.imgur.com/DQuLYzz.png) when he finally got to place them via mapthing.
Specifically, I think the problem this was caused by was a straight up copypaste of the "height" attribute of MT_SPIKE into the "radius" field of MT_WALLSPIKE, when it should technically be half that to match the intended collision box. Some other stuff also got adjusted to match when I solved this.
FF's happy now - they work well enough to place, even if their [collision is a little wonky](http://i.imgur.com/o2rG0tV.gifv), but that's an MF_PAPERCOLLISION issue and more than a quick fix to be done in an evening.
Nab ```<root>/!LatestSRB2Files/srb2win_branch_spikefix.exe``` to test.
UPDATE:
* Support for MF_PAPERCOLLISION tmhitthing in P_SlideMove.
* Knockback for wallspikes is much more consistent, fixing FF's gif as posted above/below.
* Optimised away a FixedDiv call when determining the slopetype of a line.
* Changed the position of the wallspike base slightly.
* Made it so NOTHING can step up onto the player outside of other players, essentially killing a bunch of springs-launch-with-players style bugs. (I was able to recreate that issue with the wallspikes until I added this, which is why I did it in this branch.)
* Wallspikes are now bustable.
* Invert the solidity mapthing flag for spikes and wallspikes.
* Improve the frame arrangement for vertical spikes' spriteset. (Requires new patch.dta)
* Make all spikes consistently do damage with DMG_SPIKE.
* (unrelated) renamed "flame stomp" to "elemental stomp" in P_HitDeathMessages.
* (unrelated) Make gas jets not make noise if they have the ambush flag, per Sphere's request.
See merge request !107
This allows saving in special cases where you can go to a map that
doesn't save, and then back to the map you saved on (see: SUBARASHII
special stages), while still preventing you from killing your lives by
repeatedly retrying in said map.
Kept lastmapsaved just in case. Can be removed if deemed unnecessary.
* Fix an issue where everyone being game overed except for one person, and then kicking that one person, meant the game over trigger was never met.
* Fix an issue where the spectator text could be overridden with a count of the number of remaining players to complete the level.
* Fixed a few glitches with spinning. (Unrelated to the branch, but exposed through new_coop testing.)
* If NONET, the Multiplayer option on the main menu leads directly to the Splitscreen menu, which also includes player 1/2 setup under this circumstance.
* A mechanism to save player name, skin and colour as defaults to config.cfg.
* Improve the code for busting vertical spikes.
* Improve the frame arrangement for vertical spikes' spriteset. (Requires new patch.dta)
* Make all spikes consistently do damage with DMG_SPIKE.
* (unrelated) renamed "flame stomp" to "elemental stomp" in P_HitDeathMessages
* Support for MF_PAPERCOLLISION tmhitthing in P_SlideMove.
* Knockback for wallspikes is much more consistent.
* Optimised away a FixedDiv call when determining the slopetype of a line.
* Changed the position of the wallspike base slightly.
* Made it so NOTHING can step up onto the player outside of other players, essentially killing a bunch of springs-launch-with-players style bugs. (I was able to recreate that issue with the wallspikes until I added this, which is why I did it in this branch.)
Or rather, I killed anything to do with the old background orb, renamed all "EMMY" stuff to "TOKEN", and of course adjusted the new S_TOKEN accordingly
You can either add FF_SPR2SUPER directly to the sprite2 arg if it's a number, or you can use the new (optional) boolean arg between sprite2 and frame, where as you'd expect false = normal and true = super.
* Crash prevention if total somehow becomes zero.
* Don't waste time multiplying and demultiplying the HUD check if it's just gonna net you the same number.
Pwease no kicky
Don't kick Tails! Also, a movement for the WRITESINT8 to prevent modification to buf if the function bails early.
This can go into Master, right? It only matters for the host, and it's explicitly only having a major effect outside of netgames.
See merge request !201
*Add CONS_Printf messages for !netgame checks
*Arg count is checked first regardless of netgame status for both kick and ban, < 2 is checked instead of == 1 just in case these weren't called from console for some stupid reason?
*Moved Command_Kick's buffer vars to within the code that actually does kicking stuff
* Hide the individual lives on the tab menu whenever they're irrelevant.
* Make co-op spectators appear on the tab menu.
* Hide one frame of spectatorhood stuff in splitscreen.
* Fix joining netgames whilst in a special stage not giving the startinglives.
* Make monitors that don't appear to do anything play the failure sound.
* Only make players flash if they're REspawning, rather than spawning into a fresh level - unless it's a NiGHTS level, in which case never.
* Make the default value of cv_playersforexit "All".
* Correct the greying out of the lives counter for cv_cooplives == "Single pool" to only be when you can't respawn.
NEW COOP BASED:
* Add "Infinite" option to cv_cooplives, inspired by SUBARASHII. Lives still exist, but are hidden from the player's view, and are prevented from falling below 1 at any cost. As a result, made the variable CV_CHEAT.
OTHER MULTIPLAYER BASED (semi-related):
* Made cv_autobalance an on/off switch which determines the allowed difference based on the number of people in the server, instead of a weird and opaque number from 0-4.
MENU BASED (not related):
* Add horizontal arrows to menu options which respond to the arrow keys.
* Make the menu arrows bob.
* Switch out the seperate arrows for combination arrows on the joystick menus.
* Minor cvar description tweaks.
* Don't attempt to respawn anybody when changing cv_coopstarposts and nobody CAN be respawned.
* Make Rings transparent (and not flash) when you're a spectator.
Lua archive value fix
This fixes some potential problems with archiving mobjinfo_t/state_t data via Lua in netgames (whether as custom mobj/player vars or by NetVars hook). Thanks to LJSonic for pointing this out to me.
See merge request !200
Apart from the fact that UnArchiveValue reads UINT16 for both anyway (which alone causes problems), but UINT8 isn't even enough to store the higher end of the object types list and definitely most of the states welp
