The change removes all direct calls to mxml from the modules that need an XML
parser (lastfm.c, pipe.c, rssscanner.c and httpd_rsp.c).
Even with the help of mxml, reading XML is hard, so a layer is added which
helps deal with stuff like whitespace and CDATA. This should make OwnTone more
resilient to any XML variations it might receive.
The changes fixes issue #1677.
Despite that it seemingly supports multithreading it picks threads that are
busy, and not sure how that can be prevented. Also libevhtp is currently
unmaintained.
Roku SoundBridge devices can be treated as dumb speakers; they offer a
texted based protocol, RCP, that allows the server to request the Roku
device to connect to the server's stream.mp3 and provides ability to
control volume
Roku devices appear as speakers automatically via mDNS announcements on
the server _roku-rcp._tcp
Replacing the antlr parsers solves multiple issues:
- Build warnings (issue #307)
- Build hacks: "-Xconversiontimeout 30000" and other Makefile magic
- Incorrect parsing of daap queries with sql wildcards (like 'tag:*tes%t_ng*')
- Infinite recursion/memory/CPU (issue #570 and #1248)
- systemd service file workarounds due to memory/CPU issue
- ANTLR3 being replaced with ANTLR4 (that doesn't support C file generation)
- Runtime dependency on antlr library
- Difficult installs of ANTLR3 on some systems (special install script)
Select use of either libspotify or librespot-c as streaming backend via config
option.
librespot-c (renamed/improved spotifyc) impl has the following:
- sync interface
- seek support
- honor bitrate config, set client and thread name
- use web access token with "streaming" scope for login
- fix issue with podcast playback
Also say goodbye to file-based Spotify login.
Includes
- Implementation in src/outputs/airplays2, type OUTPUT_TYPE_AIRPLAY
- Homekit pairing, both normal (with PIN) and transient
- New session startup sequence, incl GET /info, SETPEERS and 2 x SETUP
- No more OPTIONS and ANNOUNCE
- Use POST /feedback for keepalive instead of SET_PARAMETERS
- Sequence dispatching instead of callback chains
- Continue despite "Bad request" to SET_PARAMETER (volume)
- Opening of event connection to receiver (reverse rtsp connection)
Still to be done
- Password authentication
- Handling of events
This adds a new settings component for user configurable options that
can be changed through the JSON API.
The settings are stored in the admin db table and not in the conf-file.
- also change how speakers are saved/retrieved from the db
- add generic authorization methods in outputs.c and player.c
- let filescanner read *.verification files (containing PIN)
- configure options to enable and disable, since libsodium is required
* [config] Added support for libav/ffmpeg dual installs
Reordered CPPFLAGS/LIBS so that library flags (includes etc) appear
ahead of prefix/global directives and the correct headers/libraries
are checked (if symbols are defined in alternative libraries).
Updated libav checks to use header declaration checks
rather than link checks (so differences between
libav/ffmpeg can be correctly found), and updated code references
to use new defines. Added support the alternate pkg-config
package names for libav. Updated ffmpeg vs libav check
to check MICRO version number (>=100 for ffmpeg). Simplified
resulting configure script by using a function to merge
CPPFLAGS variables.
* [config] Use FORKED_OPTS prefix for libevent_pthreads option
* [config] Add --with-libav option to select libav even if ffmpeg present
