- rename from "libsndio" to "sndio"
- remove support for "frame tick" synchronization; nothing uses it
- as with many other sndio backends, the audio device block size is the
buffer-write-size the application wants and there are 2 blocks per
device buffer
- if SDL will resample, increase the audio block and buffer size by
the same ratio, so that device latency (or how much *time* the
application has between write()s to not let the buffer underrun) does
not change
- allow applications to disable conversions
tested with almost every port that uses this code (exceptions being some
games that require non-free game data) on a device most likely to be
affected by the changes (azalia(4) that only does 44.1 or 48 kHz s16)
unoverriding conversion disabling exposes brokenness in a few other
ports, those will be fixed soon
is not set elsewhere.
- Add disable option for PulseAudio so support is not built if it happens
to be installed.
- Enable the use of YASM to build the i386 assembly blit code.
ok pvalchev@
of trying to dlopen() it. also fix a couple bugs in the artsd
output module: when artsd is "suspended", it only means that
it it not accessing the soundcard at the moment, not that it
will not accept data. fix output latency by setting parameters
that actually are settable, instead of trying to set read-only(!)
parameters.
- allow the default audio output method to be chosen with the
SDL_AUDIO_DRIVER environment variable. choices are 'arts',
'esd' and 'audio'. 'audio' selects audio(4) output, and is
the default if SDL_AUDIO_DRIVER is not set.
ok pvalchev@
Bugfix release, notably:
- Fix a fullscreen crash with ffmpeg, etc. due to expecting memory to
be executable by default - use mprotect() - Reported by deanna@
- Worked around BadAlloc error when using XVideo on the XFree86 Intel
- Integrated Graphics driver
- Fixed use of sdl-config arguments in sdl.m4
note: people running with non-default FLAVOR(s) will need to update their
package manually as "pkg_add -u" won't be able to cope with this change
discussed with and ok steven@, "theorical ok" from pvalchev@
default, so now we:
* add an altivec FLAVOR
* mark is BROKEN (so that --disable-altivec is used)
Note that this is not a "set and forget" workaround, we will try fixing
this segfault but for now we need a working SDL for powerpc.
discussed with and ok pvalchev@
- sync patches with update-patches
- bump PKGNAME
the sun interface keeps fewer samples in the play buffer, and with
less fluctuation than the OpenBSD interface. results in better
audio-video synchronization in e.g. ffplay and gnash
go for it -pval
X11 >= 10, and bump those major numbers accordingly, so that bug-reports
will get easier to process.
(artificial `flag day' as discussed with deraadt@)
at least some macppc scenarios; discovered by ajacoutot
since it's only a few hour window of breakage, be careful reverting
things if you updated during that time
- sdl-config no longer provides -I${LOCALBASE}/include.
- add an `aa' flavor (ASCII art display).
- GNU make is not needed.
- don't use nasm with gcc 3.
- don't install a SEGV handler.
- prevent the joystick driver from picking up usb keyboards.
ok pvalchev@
