The previous function could only be effectively run one because
it changes the value of the global 'joytype' to something other
than AUTO, which is what this function looks for to re-query
the joysticks.
Second, if this function is run more than once, then it clobbers
the prior set number of joysticks back to zero (without
re-requerying for the actual quantity), rendering the joysticks
unuseable.
Finally, this function depends on several SDL calls to the joystick
subsystem, but fails to check or initialize it.
The commit:
- Checks for an initializes the joystick subsystem before using it
- Always re-queries the current state of the joystick (re-runnable)
- Simplifies the logic to determine useability (retains criteria)
- Conservatively writes the joysticks quantities at the end
There's no reason to use more than 32 bit for this parameter.
Also, limit scope of certain variables as much as possible without
bigger refactorizations.
There's no reason to use more than 32 bit for this parameter; source
values are 16 bit, and they occassionally shifted, but always to values
fitting in 32 bit.
In an earlier change, I removed appending newline outside of batch mode
in DOS shell code - that made DOSBox behave less like MS-DOS and more
like modern shells, that do not try to compensate for buggy
applications.
However, we should recognize that DOSBox (unlike e.g. FreeDOS) is designed
to run legacy applications, which might make assumptions about DOS
implementation. Some examples:
- PC Player Benchmark assumes, that help commands are displayed exactly
at 80x25 terminal and formats the output to fill the whole screen
(scrolling past DOS4GW messages).
- Quake and other ID games print shareware information on exit, but do
it via a direct memory dump (not interrupts to print DOS text), and
follow up with setting cursor exactly at line 22 (which is partly
written already), expecting shell to inject newline.
- PCC Compiler prints status message on exit without newline, depending
on MS-DOS shell behaviour.
- TEXTUTIL set of external commands do not print nothing to standard
output, and are designed to clear the screen, therefore writing a
newline after .COM commands would be a mistake.
Therefore we want to inject this newline, but not in every case.
New implementation reuses a static variable used by Program base class
(for purpose of translating UNIX newlines to DOS newlines) for detection
if it's appropriate to inject an additional newline or not.
Injecting the newline happens in function displaying the DOS prompt (so
we don't need to write additonal logic for separately handling batch
mode). When starting a non-COM, non-internal command the static variable
is set to the state indicating that next DOS prompt should inject the
newline.
Fixes: #208
Both new SVN commits are irrelevant to dosbox-staging:
- overlay output does not exist any more since SDL2
- we already switched to OpenGL on all OSes, not only macOS
Changes were removed during conflict resolution. Keeping empty merge
commit to aid in future imports.
Clang memory sanitizer needs all linked libraries (including STL) to be
compiled with msan option, otherwise analysis results are practically
unusable.
GCC undefined behaviour sanitizer analysis is already covered by build
running UASAN (undefined behaviour and address sanitizers).
In the current code, we assess the --disable-opus-cdda argument,
and if it's not present we run pkg-config to get the Opus CFLAGS
and LIBS. If it finds them, it sets the OPUSFILE_CFLAGS and
OPUSFILE_LIBS variables. We finally check for the presense of
one of these before actually enabling Opus.
However, if the user directly sets OPUSFILE_LIBS, then this final
check will become true regardless if the user passed --disable-opus-cdda
to configure.
The obvious fix is to move the final Opus check inside the
prior checks (for example, moving AM_CONDITIONAL(USE_OPUS) inside
AS_IF(..), however that results in the OPUS_USE conditional not
existing if the AS_IF test fails its check.
Therefore, we set a new variable HAVE_OPUS, if the pkg-config
check passes, and look for that instead of OPUSFILE_LIBS.
(Of course, the user could export HAVE_OPUS=yes in their
environment as well, but at that point they're deliberately
circumenting configure, in which case they can have at it!).
The new output type `texturepp' was added, which implements
pixel-perfect scaling using SDL's hardware-accelerated texture output.
In pixel-pefect mode, each original pixel is displayed as a rectangle m
by n pixels, so that m:n yields a reasonably good approximation of the
pixel aspect ratio (PAR) of the emulated graphical mode while using as
much screen space as possible. The balance between the precision of
aspect ratio and the utilisation of screen space is specified as the
`parweight' parameter to pp_getscale() and is currently hard-coded in
sdlmain.cpp.
This implementation emulatates pixel-perfect mode as a special case of
nearest-neighbor interpolation when the horisontal and vertical scaling
factors are integers.
Our macOS release builds are limited to Catalina only, and this OS
officially works only on Ivy Bridge (2012) or newer microarchitecture.
Point GCC to target Nehalen (2008) instead - this should cover also all
possible Hackintoshes or virtual machines.
Pass --disable-network to disable modem and ipx features during
configuration step (and avoid linking SDL2_net without actually removing
it from the system).
The hashest generated by xxHash have changed between v0.7.2 and v0.7.3,
therefore the hashes in prior-genreated lookup tables will no longer
be valid. By bumping the file's identifier, the fast-seek code will
reject these older files and regenerate new ones.
