* Add more descriptions for CVARs
* Fixed a small typo
* Another small typo
* Added the suggestions from @Shpoike and @eukara along with a little more cleanup
* Workaround for the OGG decoding issue. 5159a80 was bad for ogg vorbis decoding. It was never ideal to begin with, but it's better than rendering the game unplayable.
* Concentrated patch, courtesy of Spike so we can still mark sounds as nopurge
* Quake II: read the skin, effects, renderfx, and solidsize fields as unsigned shorts.
* Add missing header definition for int MSG_ReadUShort();
* Rename MSG_ReadUShort() to MSG_ReadUInt16() as per @Shpoike's wishes
* solidsize needs to be read in via MSG_ReadSize16 because it does protocol conversions (@Shpoike)
* Q2UFRAME16 and Q2USOUND should be read unsigned as well, as it matches the internal data within FTE (@Shpoike)
* Remove all outdated Visual Studio project files
* remove gas assembler, I doubt we need this 20 years later
* Remove final visual studio junk, this time for NPAPI
Forgot to mention previous commits also removed "libs/dxsdk*" and fixes#193
* Removed libs as requested by @eukara and @Shpoike
* Remove the nacl dir
* Remove nacl from ifdef checks shared with other platforms
* Remove nacl from build bash scripts
* More removal of nacl from bash scripts
* Remove nacl-specific code from the engine
* More removal of nacl-specific code from the engine
* remove nacl-specific code from the Quake 3 plugin
* Remove nacl from readme (also seems to have force LF line endings)
* Remove nacl mention from browser.txt
* Remove nacl from Makefile
* Remove nacl from dotnet2005 solution file (lol)
* Correcting an endif, this did not end nacl
* Clarification for NaCL no longer needed
* Left small todo for myself
* Partially undo f44daef091bcc5917b0ddb2cbe41ebdd9738f770
the if preproc checking for `_DEBUG` was incorrectly removed
We should now support dtls1.2 under win10 (otherwise win7+8 is still stuck with 1.0).
SSQC can now query client certificate info via infokey - *cert_sha1 or *cert_dn
Server addresses can be postfixed with eg ip:port?fp=BASE64 to provide a fingerprint to verify the server without depending on cert authorities.
I did this because I wanted to fly around maps using a Spaceball
4000FLX, or any other 6DOF controller. These fixes help it work.
Various fixes to joystick support:
- joyaxiscallback() used strtol() to check to see if the supplied
string was an integer, then didn't assign the parsed integer to
the cvar.
- Wrong multiplier for left/turnleft values.
- Delete `axismap[]` from J_JoystickAxis(). It was causing problems,
and smells like it was trying to do what the cvars
joyadvaxis[xyzruv] are doing now.
- Fix compiler error by adding case statements for:
SDL_SENSOR_ACCEL_L
SDL_SENSOR_ACCEL_R
SDL_SENSOR_GYRO_L
SDL_SENSOR_GYRO_R
New cvar: "joyonly".
"Joystick" axes are typically return-to-center affairs; their deflection
values are therefore reported as -MAX - MAX, with zero in the center.
"Game controllers" are similar, but also often have analog left and
right "triggers" which are reported as 0 - MAX, with zero at one end
(fully released to fully depressed).
Unfortunately, SDL will try its darndest to make a joystick look like a
"game controller." It does this by reinterpreting certain of the axes
to report the range 0 - MAX, as if they were triggers. This is not a
thing to do with 6DOF controllers, where all axes are return-to-center.
While it may be remotely possible to put together an SDL2 controller
mapping that reports -MAX - MAX on all axes, for me it was simpler to
hack on FTEQW.
Coupled with that is FTEQW's giving preference to "game controllers,"
i.e. if SDL_IsGameController() returns true, FTEQW will treat it as one.
"joyonly" is a boolean cvar. If true, FTE will ignore "game
controllers" and treat everything as a joystick. The default is false.
"joyonly" must be set at startup to be effective, when the controllers
are being enumerated by SDL.
