New subject: Fix for distortions at file split points [was: Re: [ANNOUNCE] VDR developer version 1.7.19]

19 Jun 2011


      VDR developer version 1.7.19 is now available at
ftp://ftp.tvdr.de/vdr/Developer/vdr-1.7.19.tar.bz2
A 'diff' against the previous version is available at
ftp://ftp.tvdr.de/vdr/Developer/vdr-1.7.18-1.7.19.diff
MD5 checksums:
1eb04ecdc2b11ab8641ebfaa2cb93f42  vdr-1.7.19.tar.bz2
db16ce7bb51e0db837ed56ef4744a67e  vdr-1.7.18-1.7.19.diff
WARNING:
========
This is a *developer* version. Even though *I* use it in my productive
environment. I strongly recommend that you only use it under controlled
conditions and for testing and debugging.
This version introduces functions to determine the "signal strength"
and "signal quality" through cDevice. If you are using a DVB card that
contains an stb0899 frontend chip (like the TT-budget S2-3200) you may
want to apply the patches from
ftp://ftp.tvdr.de/vdr/Developer/Driver-Patches
to the LinuxDVB driver source in order to receive useful results from
that frontend.
Since apparently the various frontend drivers return different maximum
values in their FE_READ_SIGNAL_STRENGTH and FE_READ_SNR functions (some
deliver a value in the range 0x0000...0xFFFF, while others return values
as "dB/10" or "dBm/10" (the latter with an offset to make the value positive,
since the parameter is unsigned), the functions cDvbTuner::GetSignalStrength()
and cDvbTuner::GetSignalQuality() use the device's "subsystem ID" to
map these values into the range 0...100, which is the normalized return
value of these functions.
Take a look at these two functions and maybe remove the comment characters
from the lines
//#define DEBUG_SIGNALSTRENGTH
//#define DEBUG_SIGNALQUALITY
in dvbdevice.c to get some debug output if your device doesn't return any
directly useful values and may have to be added appropriately to the
'switch (subsystemId)' statement.
The channel display of the 'sttng' skin uses these values to implement
a signal strength/quality display.
The changes since version 1.7.18:
- Fixed cString's operator=(const char *String) in case the given string is the
   same as the existing one (thanks to Dirk Leber).
- Avoiding a gcc 4.6 compiler error in the skincurses plugin (thanks to Tobias Grimm).
- TsGetPayload() now checks if there actually is a payload in the given TS packet
   (reported by Dirk Leber).
- Now sorting the source file names in the call to xgettext, to make sure the results
   are not dependent on the sequence of the files.
   Plugin authors may want to change the line containing the xgettext call in their
   Makefile accordingly by changing "$^" to "`ls $^`".
- The primary device is now only avoided for recording if it is an old SD full
   featured card. This is done through the new function cDevice::AvoidRecording().
- Subtitle PIDs are now also decrypted (thanks to Reinhard Nissl).
- Fixed a possible race condition in cDiseqc::Execute() (reported by Marco Göbenich).
   The return value of cDiseqcs::Get() is now const, so plugin authors may need to
   adjust their code if they use this function.
- The new functions cDevice::SignalStrength() and cDevice::SignalQuality() can be
   used to determine the signal strength and quality of a given device (thanks to
   Rolf Ahrenberg for some input on how to use BER and UNC values to generate a
   "quality" value).
- The 'sttng' skin now displays two colored bars at the bottom of the channel display,
   indicating the strength (upper bar) and quality (lower bar) of the received signal.
   The number to the left of these bars indicates the actual device the current
   channel is being received with.
- Fixed detecting frames in case the Picture Start Code or Access Unit Delimiter
   extends over TS packet boundaries (reported by Johan Andersson).
   In order to fix this, the semantics of cFrameDetector had to be changed a little.
   See cRecorder::Action() and cIndexFileGenerator::Action() on how to use the new
   cFrameDetector::NewPayload() function.
- The frame detector now only starts collecting PTS values after it has seen the
   first I-frame, otherwise it might get MaxPtsValues values and stop analyzing
   even though the incoming data is still garbage (reported by Derek Kelly).
- The info file of a recording is now only overwritten with a new fps value if
   that new value is not the default value (thanks to Derek Kelly for reporting a
   problem with the fps value being overwritten in case a recording was interrupted
   and resumed, and the fps value could not be determined after resuming recording).
- The initial channel is now stored by the channel ID in the setup.conf file, in
   order to avoid problems in case channels are reordered or deleted (reported by
   Lars Bläser).
- Added support for "content identifier descriptor" and "default authority descriptor"
   to 'libsi' (thanks to Dave Pickles).
Have fun!
Klaus