This is highly experimental. It's defers most checkpointing
to a separate thread to see if we get better throughput and
less choppiness when applying effects.
This reenables synchronous mode by default. However, for processing
where a power cut or crash can be tolerated, synchronous and journaling
are disabled. Once the processing is complete, they are reenabled.
Types of processing that are like this are "Save As", "Backup Project",
and "Vacuuming". They all write to a separate project file while
running, so the real project file is safe.
Unfortunately, effects are back to be slow and sluggish.
I believe I've address all of the weird file corruption issues and
I'll continue to continue testing for these.
It now uses VACUUM INTO instead of the SQLite backup API
in hopes that the copies will be smaller. And VACUUM INTO
is "supposed" to be faster, but time will tell. It's easy
to put the backup API usage back in.
This also fixes a bit I missed with redoing the orphan block
handling that was reported by Paul.
And finally, it renames the AutoRecovery.cpp/.h files and AutoSaveFile
class to ProjectSerializer since the AutoSaveFile class is being
used for regular project documents now and it doesn't write to a
file anymore.
If anyone has a better idea for a name other than ProjectSerializer
feel free to change it. I hate naming things.
Preliminary tests show it to be a bit faster than the default
4KB. For a simple example, generate 2-hour chirp dropped from
11 seconds to 7 seconds. Not a lot, but...
If you try to build local SBSMS with CMake for i686, you get many error messages like this one:
audacity/lib-src/sbsms/src/fft.h:346:29: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
The solution is to add the option provided by SSE_FLAGS.
So, I replicated in this PR the same fix made for local soundtouch and LAME library.
The last commit resolved this requirement. I'd really like to
remove ALL of the old dynamic loading code from ExportMP3 as well,
but a task for another day...
* [CMAKE] Portaudio: do not activate OSS on Windows
I had a strange error when building the local portaudio library with MinGW under msys2.
The error was caused by the file sys/soundcard.h that was found somewhere in the path, because use_pa_oss is always activated regardless the platform.
So, in my opinion it is better to not activate this option if the platform is Windows.
Afterall, it is useless in this case.
* [CMAKE] Portaudio: move also remaining OSS code
As you can see from the original CMakeLists.txt into the sources of Portaudio, the file pa_x86_plain_converters.c can be compiled only with Visual C++, so it should not be include if there is not the Microsoft Compiler.
After compiling Audacity with MinGW, a strange message appeared on the console after the executable was launched:
# ./Audacity.exe
Usage: C:\msys64\home\Carlo\a\bin\Audacity.exe [OPTION]... INPUT...
Validate RDF data
-h Display this help and exit
-l Print errors on a single line.
-v Display version information and exit
Validate RDF data. This is a simple validator which checks
that all used properties are actually defined. It does not do
any fancy file retrieval, the files passed on the command line
are the only data that is read. In other words, you must pass
the definition of all vocabularies used on the command line.
The problem happens because the file lv2/sord/src/sord_validate.c contains a main() function that was called by crt startup code instead of the one in the main sources of Audacity. The solution is to not include this unneeded file.
If you try to build local LAME with CMake for i686, you get many error messages like this one:
C:/cygwin64/home/Carlo/audacity/lib-src/libmp3lame/vector/xmm_quantize_sub.c: In function 'init_xrpow_core_sse':
C:/cygwin64/home/Carlo/audacity/lib-src/libmp3lame/vector/xmm_quantize_sub.c:72:18: warning: SSE vector return without SSE enabled changes the ABI [-Wpsabi]
The solution is to add the option provided by SSE_FLAGS.
So, I replicated in this PR the same fix made for local soundtouch library.
Cliff noticed that the CMake git ident was not being updated when
pulling new changes. This was because it was getting captured at
configuration time and, even if you pull more changes, the CMake
configuration may not be redone automatically if the build files
weren't also changed.
So, this adds a new target to get the information at build time
instead.
