LogoopenSUSE Build Service > Projects
Sign Up | Log In

View File mkclean.changes of Package mkclean (Project home:Simmphonie)

Mon Apr 13 11:13:24 UTC 2015 - t.gruner@katodev.de

- Initital Release

version 0.8.7:
    - handle non-latin command-line characters on windows

version 0.8.6:
    - correct DisplayUnit that appears to be in pixels but marked as Display Aspect Ratio (DAR)

version 0.8.5:
    - fix the recompression of the CodecPrivate through successive mkclean calls

version 0.8.4:
    - fix the keyframe detection for WebM alt-ref frames (0 duration = non keyframe)

version 0.8.3:
    - fix the Segment duration when using --timecodescale
    - fix the progress display on segments with no size set (like live streams)
    - add -alt-3d to pack video frames by 2 when left/right eyes are coded sequentially in time (half duration frames)

version 0.8.2:
    - add support for Codec Private zlib compression with --optimize (for Vorbis)
    - add support for zlib compression of A_PCM and V_UNCOMPRESS tracks when --optimize is set
    - fix a bogus aspect ratio fix
    - use the generated semantic for libmatroska2 (no more mismatch with the specs)

version 0.8.1:
    - turn old StereoMode values into the Matroska v3 ones
    - remove the ClusterPosition field in WebM files as it's not supported

version 0.8.0:
    - add support for Matroska v3
    - fix some odd case where some file other than the bogus output file would be deleted after an error
    - add more padding after the SeekHead to make room for links to chapters/tags/attachments/others and/or editing
    - add a link to the first Cluster in the Seek Head

version 0.7.3:
    - add --optimize_nv to use header stripping where possible except video tracks (for some hardware)

version 0.7.2:
    - discard --optimize for WebM

version 0.7.1:
    - fix the regression test check on large files
    - fix the cues when blocks could not be read and written in the output
    - fix the timecodes when --optimize and --timecodescale are used
    - fix in libebml2 not to read data too far in a bogus EBML master
    - the regression tests pass under Linux too
    - add a --quiet option for the regression tests to disable mkvalidator error display

version 0.7.0:
    - fix an incorrect element size computing when invalid data was read
    - allow StereoMode for WebM
    - adjust the timecodescale automatically if it's too small
    - fix AC-3 duration detection issues when remuxing
    - better handling of default/mandatory values
    - internal code shuffling
    - make the regression test tool compile under UNIX

version 0.6.1:
    - add a progress phase during the cue creation as it's sometimes slow
    - safer handling of broken files
    - fix EBML reading on big endian processors (PowerPC)
    - add a regression test tool (needs mkvalidator too)
    - fix a wrong warning on files with only 1 Cluster
    - fix some errors when the source mkv was truncated at the end

version 0.6.0:
    - fix a size update issue with the new "smarter" system
    - don't delete the mkclean binary if it's run with no parameter
    - put the attachments before the Cue entries (index)
    - put cover.jpg/cover.png at the front of attachments
    - support bz2 and lzo1x track reading (transformed to zlib when writing)
    - make sure cluster timecodes are always incrementing
    - add --regression to output files that can used for comparison in regression tests
    - fix OS X building (gcc_osx_x86 and xcode_x86)
    - fix float handling on some systems (thanks Monty)

version 0.5.7:
    - fix a problem with attachments in safe mode (--unsafe not set)
    - put the attachments at the front as it usually makes more sense (fonts or cover art)

version 0.5.6:
    - fix a major memory leak that could result in unexpected errors with large file
    - fix the UNIX source package

version 0.5.5:
    - fix a timecode bug when --timecodescale was specified on input
    - improve the recovery of damaged files
    - fix command line errors when --doctype x was used as the last param
    - delete the output file in case the output file didn't finish correctly
    - add a timecode offset to files created with --live as live streams usually don't start at 0
    - write some junk before the first cluster in --live mode, like a dumb server would do
    - fix the PrevSize computing in the --live mode
    - fix cleaning of files that had infinite sizes (live streams)

version 0.5.4:
    - fix a potential use of free'd memory when using the --optimize mode with --remux

version 0.5.3:
    - write CRC-32 elements on key parts for better file safety (not in unsafe mode)
    - don't write a Cue if there is only one Cluster

version 0.5.2:
    - fix the use of <input> <output> file use in the command line
    - don't allow source video tracks with 0 pixel values
    - fix a bug/crash when moving a whole Block to the next Cluster
    - fix a crash when handling files with CRC (like AviMux GUI files)

version 0.5.1:
    - fix an issue when fixing aspect ratio that only had one value (width or height) set

version 0.5.0:
    - prepend the output filename with "clean." if no output filename is specified
    - fix a bug when header compression was used but not lacing
    - add a --no-optimize option to forbid track compression
    - add a mkWDclean tool to remove the track compression from the output files

version 0.4.6:
    - remove bad cropping values
    - fix some potential issues when EBML lacing and header stripping was combined
    - support zlib compressed input and output tracks
    - don't try to strip headers of track with zlib compression or encryption
    - force zlib compression for a set of codecs (S_USF, S_VOBSUB, S_HDMV/PGS, B_VOBBTN) in --optimize mode

version 0.4.5:
    - fix --remux when BlockGroup was encountered (rather than SimpleBlock)
    - fix --optimize on video tracks

version 0.4.4:
    - fix a possible infinite loop when only --optimize was used
    - fix a potential timecode issue at the cluster boundaries

version 0.4.3:
    - fix a bug in the segment size when a placeholder was added for later tag reference
    - reject files that don't have an EBML header
    - add --help and --version CLI options
    - put the timecodescale first in the segment info, put the duration just after
    - fix the amount of phases displayed

version 0.4.2:
    - use DAR aspect ratio when pixels are not accurate enough
    - bootstrap.sh should now call the correct coremake config depending on the host platform
    - rename boostrap.sh to configure
    - avoid crashing with files that have no TrackInfo
    - read and skip custome elements rather than discarding (could result in corruption)
    - avoid a possible output corruption when the Segment size is much smaller than the original
    - fix the AppWriting string when there was no library in the source file
    - don't keep elements that are not valid for the output profile
    - check that track attachment links point to valid attachments and remove the invalid ones

version 0.4.1:
    - fix a possible data corruption when optimizing blocks
    - add a --quiet option to only output errors
    - keep the discardable flag as-is when reclustering
    - fix the --timecodescale handling

version 0.4.0:
    - allow using compressed headers (with --optimize) automatically picking the best size
    - delete CRC-32 elements until we can update them correctly
    - display the 64 bits integers correctly on all platforms
    - write the Cluster PrevSize and position in the Segment by default (can be disabled with --unsafe)
    - reserve a place after the SeekHead for a reference to Tags when they are not there (for easy editing)
    - make sure audio/subtitles frames are all keyframes (no known codec so far are not)
    - make sure the lacing flag in the track info matches what is used in the file
    - clean Haali's style aspect ratio (16/9 instead of the pixel values)
    - remove values that are the same as their (indirect) default values (like DisplayWidth/DisplayHeight and OutputSamplingFreq)
    - put the TrackNumber, TrackType and CodecId at the front of the Track elements

version 0.3.1:
    - output an error when the remuxing is not possible due to a low TimecodeScale (thanks to Ben Danper)
    - add a --timecodescale <v> option to force the global timecodescale

version 0.3.0:
    - always put the matching audio frames before the video frames
    - do a "relacing" when remuxing to put audio frames close to the corresponding video keyframes (not all audio codecs are supported)
      for Vorbis (via Tremor), MP3, AC3, E-AC3, AAC, DTS

version 0.2.5:
    - make sure only V_VP8 and A_VORBIS tracks are kept for webm
    - disable writing webm v1 (--doctype 3)
    - fix some default values in chapters
    - fix reading of elements with unknown/infinite sizes

version 0.2.4:
    - in --live mode the Cluster also have an infinite/unknown size
    - remove the Cluster duration from the --live mode

version 0.2.3:
    - add a --live option to create ouptut files that look like a live stream (not seekable)
    - do not write ClusterPosition and PrevSize (until we support it)
    - do not silently force a doctype change, but tell the user the CLI to use

version 0.2.2:
    - unobfuscate the 'webm' and 'V_VP8' strings
    - fix a crash when reading Sorenson Squeeze files (Ebml Void in the header)

version 0.2.1:
    - add an option to change the DocType name/version (no validity check)
    - put the (re)muxing date in the file (or if there was no date)

version 0.2.0:
    - add an option to remux files with better Cluster boundaries
    - enforce mandatory elements are always (virtually) there
    - remove Master elements that are missing mandatory elements and are not unique in the context

version 0.1.5:
    - create Cues from scratch
    - add a "--keep-cues" option to keep the original Cues instead of creating from scratch

version 0.1.4:
    - Allow files with no Track/Clusters (a file with only "logical" chapters is possible)

version 0.1.3:
    - By default build with support for large files (>4GB)
    - Sort the Cue entries
    - Optimize the Cue/Block linking speed by starting from, the previous matching Cluster found

version 0.1.2:
    - Fix the wrong EBML header in release builds (doh!)

version 0.1.1:
    - Fix the Seek Head computing mismatch for large files
    - Add a progress info during the cleaning (in 3 phases)

version 0.1.0:
    - Initial public release