SonarLint reports the latter to be better:

std::scoped_lock basically provides the same feature as std::lock_guard,
but is more generic: It can lock several mutexes at the same time, with a
deadlock prevention mechanism (see {rule:cpp:S5524}). The equivalent code
to perform simultaneous locking with std::lock_guard is significantly more
complex. Therefore, it is simpler to use std::scoped_lock all the time,
even when locking only one mutex (there will be no performance impact).
Signed-off-by: Rosen Penev <rosenp@gmail.com>

If UpdateStreamTag() gets called while an initial seek is pending, the
result will never be submitted to a MusicChunk.  By avoiding the
UpdateStreamTag() call in that case (by moving UpdateStreamTag() to
after the PrepareInitialSeek() check), the song_tag is preserved until
UpdateStreamTag() is called again from SubmitData().

This fixes missing tags in the "httpd" output.

Closes https://github.com/MusicPlayerDaemon/MPD/issues/1137

Found with -Wdouble-promotion
Signed-off-by: Rosen Penev <rosenp@gmail.com>

When the client wants to seek, but the decoder has already finished
decoding the current song, the player restarts the decoder with an
initial seek at the new position.  When this initial seek fails, MPD
pretends nothing has happened and plays this song from the start.

With this new flag, a restarted decoder marks the initial seek as
"essential" and fails the decoder if that seek fails.

Closes https://github.com/MusicPlayerDaemon/MPD/issues/895

Before the advent of io_uring (commit dae8da70), this didn't
matter, because the `FileInputStream` never called this.  But
`UringInputStream` is derived from `AsyncInputStream`, and needs the
handler to signal completion.

Closes https://github.com/MusicPlayerDaemon/MPD/issues/898

Found with -Wdouble-promotion
Signed-off-by: Rosen Penev <rosenp@gmail.com>

Boost does not seem to offer an overload for lrint.
Signed-off-by: Rosen Penev <rosenp@gmail.com>

The former was deprecated with C++14.

According to the C++11 and C++17 standards, both files are identical.
Signed-off-by: Rosen Penev <rosenp@gmail.com>

This optimization is useless because sane pthread_cond_signal()
implementations check the number of waiters and do not invoke a system
call if there are none.

Just like std::condition_variable, which however has no way to specify
the std::mutex directly.

If the DecoderThread gets woken up by a STOP command, this should be
detected as early as possible.

Let the constructor and destructor do this.  This means that all users
have to be converted to allocate PcmConvert dynamically.

Since we switched from autotools to Meson in commit
94592c14, we don't need to include
`config.h` early to properly enable large file support.  Meson passes
the required macros on the compiler command line instead of defining
them in `config.h`.

This means we can include `config.h` at any time, whenever we want to
check its macros, and there are no ordering constraints.

This requires re-adding the reference to struct DecoderControl, which
was removed recently by commit
9f14e7a9

Make all uses of MusicChunk safe.

This tiny amount of overhead allows omitting the MusicBuffer in
Clear().

This adds a bit of overhead, but also adds flexibility to the API,
because arbitrary triggers may be invoked from that virtual method
implementation, not just Cond::signal().

The motivation for this is to make the handlers more dynamic, for the
upcoming buffering class utilizing ProxyInputStream.