summaryrefslogtreecommitdiff
path: root/libsi/section.c
AgeCommit message (Collapse)Author
2004-02-29Version 1.3.5vdr-1.3.5Klaus Schmidinger
- Fixed reading the EPG preferred language parameter from 'setup.conf'. - Fixed switching to a visible programme in case the current channel has neither a video nor an audio PID. - Fixed editing channels (SID now range checked) and creating new channels (NID, TID and RID are now set to 0). - Fixed transponder handling to make it work with satellites that provide two transponders on the same frequency, with different polarization, like Hispasat at S30.0W (thanks to Thomas Bergwinkl for pointing this out). See man vdr(5) for details about the enhanced channel ID format. - Since there appears to be no general solution for the UPT error yet, a recording now initiates an "emergency exit" if the number of UPT errors during one recording exceeds 10 (suggested by Gregoire Favre). Since the UPT error doesn't happen on my system, this has not been explicitly tested. The "preliminary fix" for the UPT error in VDR/dvbdevice.c has been disabled by default, since it makes channel switching unpleasently slow. If you want to have that workaround back, you can uncomment the line //#define WAIT_FOR_LOCK_AFTER_TUNING 1 in VDR/dvbdevice.c. - Adapted the 'sky' plugin to use the actual channel IDs, and to fetch EPG data from www.bleb.org. - Limited automatic retuning to devices that actually provide the transponder (necessary for the 'sky' plugin). - Fixed handling receivers in the 'sky' plugin, so that a recording on the same channel won't interrupt an ongoing Transfer Mode. - Added subtable ID and TSDT handling to 'libsi' (thanks to Marcel Wiesweg). - Fixed some Russian OSD texts (thanks to Vyacheslav Dikonov). - Added the 'running status' to the EPG events. This is necessary for implementing the VPS function for recording. - Removed the obsolete 'present' and 'following' handling from the EPG data. - The EPG data is now always kept sorted chronologically in the internal data structures. This also means that any EPG data retrieved through the SVRDP command LSTE is guaranteed to be sorted by start time. - Now using the 'running status' in the channel display, so that a programme that has an end time that is before the current time, but is still running, will still be shown in the display (provided the broadcasters handle the 'running status' flag correctly). This also applies to programmes that have a start time that is in the future, but are already running. - Implemented an "EPG linger time", which can be set to have older EPG information still displayed in the "Schedule" menu (thanks to Jaakko Hyvätti). - Added PDCDescriptor handling to 'libsi'. - Implemented handling the VPS timestamps (aka "Programme Identification Label") for full VPS support for timers (provided the tv stations actually broadcast this information). The VPS time is displayed in the event info page if it exists and is different than the event's start time. - Extended the SVDRP command LSTE to allow limiting the listed data to a given channel, the present or following events, or events at a given time (thanks to Thomas Heiligenmann). - Fixed a typo in libsi/si.h (thanks to Stéphane Esté-Gracias). - Timers can now be set to use the VPS information to control recording a programme. The new setup options "Recording/Use VPS" and "Recording/VPS margin", as well as the "VPS" option in the individual timers, can be used to control this feature (see MANUAL for details). Note that this feature will certainly need a lot of testing before it can be called "safe"! - The "Schedule" and "What's on now/next?" menus now have an additional column which displays information on whether there is a timer defined for an event, whether an event has a VPS time that's different than its start time, and whether an event is currently running (see MANUAL under "The "Schedule" Menu" for details).
2004-01-04Version 1.3.0vdr-1.3.0Klaus Schmidinger
- Changed thread handling to make it work with NPTL ("Native Posix Thread Library"). Thanks to Jon Burgess, Andreas Schultz, Werner Fink and Stefan Huelswitt. - The cThread class now accepts a 'Description' parameter, which is used to log the beginning and end of the thread, together with its process and thread id. For descriptions that need additional parameters you can use the function cThread::SetDescription(), which accepts 'printf()' like arguments. Existing plugins that use threads should be changed to use this functionality instead of explicit 'dsyslog()' calls inside their Action() function in order to support logging the thread ids. - Added "Slovak Link" and "Czech Link" to 'ca.conf' (thanks to Emil Petersky). However, 'ca.conf' is now pretty much obsolete due to the automatic CA handling. - Mutexes are now created with PTHREAD_MUTEX_ERRORCHECK_NP, which makes the 'lockingTid' stuff obsolete (thanks to Stefan Huelswitt). - Changed font handling to allow language specific character sets. - Adopted the small font character set from the "Elchi" patch (originally provided by Alessio Sangalli). - Greek language texts now use iso8859-7 character set (thanks to Dimitrios Dimitrakos). - Rearranged section data handling, so that the actual data handling can be done separately, even from within plugins. - The EPG data structures have been moved from eit.[hc] to epg.[hc] and have been adapted to the general VDR coding style. Plugins that use these data structures may need to change some function names (which should be obvious). The name 'subtitle' has been changed to 'shortText' to avoid clashes with actual subtitles that are part of a movie. The name 'extendedDescription' has been shortened to 'description'. - Replaced 'libdtv' with 'libsi' (thanks to Marcel Wiesweg), which is thread safe and can be used by multiple section filters simultaneously. - Added 'cRwLock' to 'thread.[hc]'. Note that all plugin Makefiles need to define _GNU_SOURCE for this to work (see the example plugin Makefiles and 'newplugin'). - Fixed a problem with crc32 in SI handling on 64bit systems (thanks to Pedro Miguel Sequeira de Justo Teixeira for reporting this one). - Fixed an alignment problem in CAM access on 64bit systems (thanks to Pedro Miguel Sequeira de Justo Teixeira for reporting this one). - Added 'StreamType' setting to CAM communication, which is important for Aston/SECA CAMs (thanks to Antonino Sergi). - Now the CA descriptors are sent to the CAM in the 'program' or 'ES level' sections, depending on where they are found in the PMT (thanks to Hans-Peter Raschke for reporting this one). This should make SkyCrypt CAMs work. - Now using the 'version number' of EPG events to avoid unnecessary work. - Channel data is now automatically derived from the DVB data stream (inspired by the 'autopid' patch from Andreas Schultz). - The current channel is now automatically re-tuned if the PIDs or other settings change. If a recording is going on on a channel that has a change in its settings, the recording will be stopped and immediately restarted to use the new channel settings. - EPG events now use the complete channel ID with NID, TID and SID. - Channel names in 'channels.conf' can now have a short form, as provided by some tv stations (see man vdr(5)). Currently channels that provide short names in addition to long ones are listed in the OSD as "short,long name", as in "RTL,RTL Television". The short names will be used explicitly later. - The Ca parameter in 'channels.conf' has been extended and now contains all the CA system ids for the given channel. When switching to a channel VDR now tests for a device that provides one of these CA system ids. The devices automatically get their supported ids from the CI handler. - The values in 'ca.conf' are currently without any real meaning. Whether or not a channel with conditional access can be received is now determined automatically by evaluating its CA descriptors and comparing them to the CA system ids provided by the installed CAM. Only the special values 1-16 are used to assign a channel to a particular device. - Increased the maximum number of possible OSD colors to 256. - Limited the line length in the EPG bugfix report, which appears to fix a buffer overflow that caused a crash when cleaning up the EPG data (at 05:00 in the morning).