- Feb 14, 2011
-
-
Jonathan Marshall authored
fixed: Scanning episodes which had a local fanart.jpg would result in the image being cached for each item.
-
elupus authored
This causes slow performance on smb:// aswell as potentially failures on any filesystem that indicate a chunksize smaller and not aligned to a normal dvd sector size (doubtfullt we have any currently).
-
Anssi Hannula authored
Extend commit 16784591 to the other direction by allowing passthrough to be stopped if necessary due to a sample rate change. Suggested by elupus.
-
Anssi Hannula authored
-
Arne Morten Kvarving authored
-
taxigps authored
-
firnsy authored
-
firnsy authored
-
- Feb 13, 2011
-
-
Jonathan Marshall authored
fixed: #10968 - autoclean library doesn't remove tvshows that no longer exist. Thanks to PhracturedBlue.
-
spiff authored
-
Jonathan Marshall authored
-
Jonathan Marshall authored
-
- Feb 12, 2011
-
-
S. Davilla authored
[vda] fixed, some encoders think 3 byte NAL sizes for avcC is valid, buzz they are not so convert to 4 byte NAL sizes when found
-
Lars Op den Kamp authored
-
Anssi Hannula authored
This avoids issues if one tries to do if (x) SAFE_x(y); else ... Now SAFE_x() behave more like regular functions, and therefore one needs to use ';' when appropriate. ';' is added where missing.
-
Anssi Hannula authored
For passthrough to work, the sample rate needs to be known. In some cases the demuxer doesn't provide the sample rate, causing us to fallback to a software codec. Retry passthrough if we finally get the sample rate during software decoding. Demuxer doesn't provide the sample rate e.g. when 1) the audio track was dynamically added (e.g. network/dvb stream), or 2) the demuxer doesn't support providing the sample rate, or 3) in ffmpeg demuxer (which actually decodes the frames from the initial tracks to get all possible information), the analyzing duration is too short for getting the detailed information for all tracks of this particular stream.
-
vdrfan authored
-
queeup authored
-
CrystalP authored
The function is already called internally by swscale when creating a swscale context. The calls would be necessary only if xbmc called the rgb conversion functions directly, which is not the case.
-
Jonathan Marshall authored
cleanup: removed effectively dead code for <system> in <coordinates> tag of skins - it basically didn't do anything.
-
- Feb 11, 2011
-
-
Anssi Hannula authored
Use the newly public higher-level libavformat probing function av_probe_input_buffer() for probing when available. This fixes several cases where files could not be played back due to too small a probe buffer. Also, lavf can automatically re-use data read by av_probe_input_buffer() for subsequent buffer reads, allowing us to drop the extra seek to the beginning after the probe when that function is used. Keep the old lower-level method for use when built against old external FFmpeg or when we want to probe for IEC 61937 or DTS specifically (dts-in-wav, ac3-in-wav, AC3/DTS audio discs).
-
Anssi Hannula authored
-
Anssi Hannula authored
FFmpeg patches from upstream that make ff_probe_input_buffer public as av_probe_input_buffer, allowing us to use it, which will make format detection more robust.
-
Anssi Hannula authored
This reverts commit 7605bca0. Removing the patches apparently didn't actually fix the issues it was supposed to fix.
-
vdrfan authored
-
Jonathan Marshall authored
-
S. Davilla authored
fixed, use @ARCH@ and let configure substitute it as the previous form is only valid AFTER Makefile.include. This fixes make clean missing libmicrohttpd
-
taxigps authored
-
- Feb 10, 2011
-
-
vdrfan authored
-
vdrfan authored
-
vdrfan authored
-
vdrfan authored
-
vdrfan authored
-
theuni authored
See: https://github.com/xbmc/xbmc-live Live may now be maintained independently from XBMC, meaning that bugfix releases (of Live itself) may be cut, and XBMC can be tagged independently. This should make development much less restricted. Acked by lcapriotti.
-
althekiller authored
Decruftify our logging class... - Remove unused functions - Remove silly dependencies - Advanced settings - Windowing system - Abstract platform descrepancies - Light refactor
-
althekiller authored
No one seems to know why we're setting the share status of the log. Not doing so doesn't seem to affect it in any way, including copying the file while XBMC is running as was hinted in the original commit (5e2ec0ca) implied. Use _wfopen instead of _wsfopen.
-
althekiller authored
-
althekiller authored
-