strlen() and strcpy() are provided by the <string.h> and <cstring>
headers (as functions in global and std namespaces, respectively).
Compilers MAY provide an implementation for either of the functions
without including the extra header but the existence of a declaration
without the header is not assured.
This issue occurs when playing HLS streaming delivered
from a server that does not support partial requests.
The issue is reproduced as follows(using Ubuntu 20.04 PC):
1. Prepare HLS example content.
$ mkdir test
$ ffmpeg -i example.flac -vn -c:a aac -b:a 128000 -f hls -hls_list_size 0 test/output.m3u8
(ffmpeg 4.2.4 is used)
2. Prepare web server without partial requests support.
(Docker version 20.10.12 and NGINX official Docker image is used)
$ docker run --name tmp-nginx-container -d nginx
$ docker cp tmp-nginx-container:/etc/nginx/conf.d/default.conf .
$ docker rm -f tmp-nginx-container
Edit default.conf and add "max_ranges 0;" to "location / {...}".
This disables partial requests support,
removes 'Accept-Ranges: bytes' header from the server response.
Then, run the server:
$ docker run --name test-nginx -v $PWD/test:/usr/share/nginx/html:ro -v $PWD/default.conf:/etc/nginx/conf.d/default.conf -d -p 8080:80 nginx
3. Setup MPD to Play the next URL.
http://address-of-the-server:8080/output.m3u8
Seeking this stream results in "exception: Not seekable".
Even if the "wave_encoder" option is disabled (and no other encoder
plugins are enabled), forcefully enable the Wave encoder (if Snapcast
is enabled).
Closes https://github.com/MusicPlayerDaemon/MPD/issues/1500
This script setup a dummy android native app folder and call ndk-gdb from it.
It needs a modification in ANDROID_NDK since ndk-gdb may attach to the wrong
pid, cf. comments in the script.
Previous versions of MPD would call SetVolume on enabled outputs before
they are ready, causing all of MPD to crash. Checking the really_enabled
flag prevents this, though it also prevents setting volume before the
player starts.
Before (with the PipeWire output):
[i] ~$ mpc clear
volume: 81% repeat: off random: off single: off consume: off
[i] ~$ systemctl --user restart mpd.service
[i] ~$ mpc volume 100
MPD error: Connection closed by the server
[i] ~ 1 $
After:
[i] ~$ # mpd is freshly started w/o anything in the queue
[i] ~$ mpc
volume:100% repeat: off random: off single: off consume: off
[i] ~$ mpc volume 80
MPD error: problems setting volume
[i] ~ 1 $ mpc
volume:100% repeat: off random: off single: off consume: off
[i] ~$
Previous versions of MPD would, on parameter change, set the PipeWire
volume before clearing the restore_volume flag, causing the call to
short circuit and do nothing. Instead, clear the flag before the call.
Max Kellermann