| Age | Commit message (Collapse) | Author |
|---|
|
In some cases a joystick axis can be "stuck" and have a large
uncentered value. These must be detected and ignored, otherwise the
axis can be chosen during calibration instead of the user's actual
desired axis.
|
|
Backspace or delete clears other input boxes; make it do the same for
keyboard, mouse and joystick inputs.
|
|
Hexen was missing the code to handle the strafe left/right buttons.
|
|
As a second fallback, if all joysticks are centered and no buttons
are pressed, look at the hats on the joystick to see if one of those
is uncentered. Bizarre as it seems, some gamepads actually present the
D-pad as a hat.
|
|
Just as some controllers have D-pads that are presented as a set of
buttons, some other controllers actually present their D-pads as
joystick hats. Add "hat axis" support where the hat number and
axis can be packed into a single integer value.
|
|
The joystick dead zone is the range within which the joystick is
interpreted as being centered. Increase this range significantly, as
some controllers (eg. the PS2 controller + USB connector) can be very
sensitive, leading to the player spinning randomly even when the stick
is centered.
|
|
Detect when we need to configure a "button axis" by falling back to
buttons when none of the joystick axes are significantly outside of
the centered range. Add extra calibration stages to get the D-pad
buttons for right and down in these cases.
|
|
Some gamepads, notably the PS3 SIXAXIS controller, provide the D-pad
not as a pair of axes, but rather as four separate buttons. Define a
special axis numbering scheme that packs two button numbers into a
single number, and allow an axis to be defined this way.
|
|
ZDoom has defined a format for Vorbis metadata comments named
LOOP_START and LOOP_END that allow the start and end points to be
defined in .ogg and .flac files for looping music. Add support for
these (they are used in Brandon Blume's SC-55 recordings).
|
|
We are free()ing the const char* a few lines later.
|
|
Clients receive the WAD and dehacked checksums of the controlling
player and are supposed to display a warning at the startup screen if
they do not match. However, reversed logic in the code that sends the
waiting data to clients meant that they were always sent their own
checksums, so the error message was never displayed.
This fixes #384.
|
|
The table of IWAD data is stored in const memory, so make all iwad_t
pointers const to fix compiler warnings.
|
|
Having multiple binaries can cause some confusion - some users try to
run chocolate-doom with hexen.wad, thinking it is supported. Add a
startup check that makes sure the user is not trying to start the game
using the wrong IWAD file for the binary being run.
This fixes #382.
|
|
Vanilla Doom's setup.exe only made reference to "joysticks" as back
then gamepads were relatively uncommon for PCs and ports for game
controllers canonically known as "joystick ports". Nowadays it's far
more likely (and ergonomic) that the player will be using a gamepad
than a joystick. Change the labelling to refer to "Gamepad/Joystick"
or "controller" instead.
|
|
When using a joystick or gamepad it's nice to be able to bring up the
menu without having to reach for the keyboard. This makes modern
gamepads more useful/usable.
|
|
Modern gamepads typically have 2-3 D-pads and joysticks. This means
that it's desirable to be able to use one joystick for turning and
another for strafing. Add another axis in addition to the current X
and y axes that performs strafe movement.
|
|
Some systems (fbcon SDL driver) get messed up if not cleanly shut
down. Make sure that we call SDL_QuitSubsystem on shutdown.
|
|
The config dumping command line option assumed that music lumps were
named like D_MYLUMP, but this is not the case for Heretic and Hexen,
where there is no D_ prefix and music lumps can have any name.
Change the logic to instead look at the contents of lumps and identify
music lumps from the MUS / MIDI header.
|
|
Some machines don't work well with 8-bit screen depths any more. It's
better to default to just using the machine's native color depth
instead. Change the default to 0 (for SDL_SetVideoMode this means "use
native color depth"), auto-adjust to native color depth on startup if
screen_bpp=0 (so that debug messages at least make sense) and document
for the config file value that a value of zero means "use native".
|
|
This matches the -nodes parameter that ipxsetup has; interpret it as
an "auto-start" parameter that launches the game when the intended
number of clients have joined the server.
|
|
Always call SDL_SetVideoMode() with bpp=0 to use the native pixel
depth of the display. This avoids problems with some systems that
don't properly support 8-bit screen modes any more. Draw into an
intermediate buffer and let SDL take care of the pixel depth
conversion for us.
|
|
Once a game type (IWAD) has been selected, remember that selection
rather than resetting to the first item in the list when the window is
closed and reopened. This fixes #169 (thanks Alexandre Xavier).
|
|
Vanilla Doom allows warp with the IDCLEV cheat up to MAP40, though it
normally crashes. Match this behavior and set MAP40 as the maximum
rather than MAP34. This fixes #181 (thanks Alexandre Xavier).
|
|
In the C standards, malloc(0) may return NULL without indicating a
failure to allocate. As values read from the MIDI file may be
arbitrary, add one to the allocated length so that we always allocated
a positive value. This fixes #165 (thanks nmain).
|
|
Sanity check the handles passed to the i_sdlsound.c API functions and
ignore requests that involve channel numbers higher than 15. This
fixes a crash if the user sets the snd_channels config variable is set
to a high value. This fixes #149 (thanks Alexandre Xavier).
|
|
Don't launch the game if a server address has not been entered; show a
message to the user instead. This fixes #121.
|
|
Using the .mp3 extension in generated music config files carries an
implicit recommendation to use that format. There are a number of
reasons to prefer other formats. The biggest is that MP3 is patent
encumbered. Furthermore, while FLAC and Ogg Vorbis both support the
Vorbis comment format that allows metadata loop point tags to be
specified.
|
|
Use appropriate directory separator for OS when generating config file
paths. When generating config files, include a header line that
indicates what is being assigned.
|
|
This adds support for high quality music packs that replace Doom's
built-in MIDI music with digital recordings. In particular this allows
recordings of the Roland SC-55 to be used in Chocolate Doom. This is
the first essential step for bug #245.
|
|
Last argument to M_StringJoin needs to be NULL.
|
|
When configuring Chocolate Strife the sound dialog can become quite
tall, and the dialog is visibly uncentered vertically. Move it up
slightly to counterbalance it.
|
|
The function would return false on lines without a '=' character, but only if that character is at least the third one on the line. :)
Closes #376
|
|
An off-by-one error in the function caused the strings to be truncated
one character too early. Change the return value check so that
negative values are also interpreted as indicating truncation; this is
the behavior of the Win32 API.
|
|
Properly handle cases where a very small string buffer is specified.
|
|
Define TXT_{StringCopy,StringConcat,snprintf,vsnprintf} as analogs of
the m_misc.c versions so that the textscreen library does not need a
dependency on the Doom code, and change all textscreen code to use
these instead of unsafe functions. This fixes #372.
|
|
Use M_snprintf() or M_StringJoin() instead where appropriate.
This fixes #371.
|
|
The Windows API has an _snprintf function that is not the same as
Unix's snprintf(): if the string is truncated then no trailing NUL
character is appended. This makes the function unsafe. Define a
replacement/wrapper called M_snprintf that works the same but always
appends a trailing NUL, for safety on Windows and other OSes that
behave like this.
Do the same thing for vsnprintf(), and update HACKING to list
snprintf/vsnprintf as forbidden functions. This fixes #375;
thanks to Quasar for pointing out the different behavior of these
functions.
|
|
|
|
|
|
The change in commit 62b5c6028.. broke the compile.
|
|
Use snprintf() in place of sprintf(). This is part of fixing #371.
|
|
Use snprintf() in place of sprintf(). This is part of fixing #371.
|
|
Use snprintf() in place of sprintf(). This is part of fixing #371.
|
|
Use snprintf() or other functions in place of sprintf(). This is part
of fixing #371.
|
|
Mistake introduced in previous commit.
|
|
As part of this, add DIR_SEPARATOR_S as a string version of the
DIR_SEPARATOR macro. Change M_TempFile() to return a string allocated
on the C heap rather than the zone heap.
This is a first step towards fixing #371.
|
|
Use M_StringCopy() instead of strncpy() in the Dehacked code.
|
|
Vanilla Hexen makes you specify the demo name to play by giving
the plain lump name, eg. heretic -playdemo mydemo to load mydemo.lmp.
It doesn't work if you specify the extension or the full file path.
As a convenience and to match the behavior of Chocolate Doom, allow
paths and extensions.
Also rework the code for other games so that they're slightly more
consistent. This fixes #301.
|
|
Vanilla Heretic makes you specify the demo name to play by giving
the plain lump name, eg. heretic -playdemo mydemo to load mydemo.lmp.
It doesn't work if you specify the extension or the full file path.
As a convenience and to match the behavior of Chocolate Doom, allow
paths and extensions.
This fixes half of #301.
|
|
Remove some remaining uses of strncpy() and use M_StringCopy() instead.
|
