| Sun, 08 Jan 2017 16:10:28 +0000 |
Chris Cannam |
Update checker with version printout
3.0-integration
|
| Fri, 06 Jan 2017 21:38:33 +0000 |
Chris Cannam |
Stub win64 deploy, + ogg testfiles
3.0-integration
|
| Fri, 06 Jan 2017 21:04:53 +0000 |
Chris Cannam |
And similar approach for the writer
3.0-integration
|
| Fri, 06 Jan 2017 20:51:47 +0000 |
Chris Cannam |
Take a different approach to using libsndfile -- the _fd function doesn't work for me in this build, so use the wchar api
3.0-integration
|
| Fri, 06 Jan 2017 16:40:11 +0000 |
Chris Cannam |
Update WAV/MP3/BZipFileDevice code to avoid using local 8-bit encoding
3.0-integration
|
| Fri, 06 Jan 2017 15:53:02 +0000 |
Chris Cannam |
Fix typo in encoded string
3.0-integration
|
| Fri, 06 Jan 2017 15:44:55 +0000 |
Chris Cannam |
More work on getting tests (especially file encoding ones) running on Windows. Various problems here to do with interaction with test filenames in Hg repos
3.0-integration
|
| Fri, 06 Jan 2017 12:11:08 +0000 |
Chris Cannam |
Add filename encoding tests
3.0-integration
|
| Fri, 06 Jan 2017 11:21:09 +0000 |
Chris Cannam |
Untabify, comment test that doesn't apply on Windows
3.0-integration
|
| Fri, 06 Jan 2017 09:50:13 +0000 |
Chris Cannam |
Untabify some more
3.0-integration
|
| Fri, 06 Jan 2017 09:43:41 +0000 |
Chris Cannam |
Improved debug output for file open troubleshooting
3.0-integration
|
| Fri, 06 Jan 2017 09:15:36 +0000 |
Chris Cannam |
Untabify
3.0-integration
|
| Thu, 05 Jan 2017 14:30:49 +0000 |
Chris Cannam |
More cache hit counts; make debug output from colour 3d plot renderer (even though it is to SVDEBUG) appear only with explicit debug flag
3.0-integration
|
| Thu, 05 Jan 2017 14:17:47 +0000 |
Chris Cannam |
Don't start in the middle when repainting the whole width while using a peak cache. It looks funny.
3.0-integration
|
| Thu, 05 Jan 2017 14:02:54 +0000 |
Chris Cannam |
Make the colour 3d plot renderer able to support more than one level of peak cache; introduce a second "peak" cache for the spectrogram layer that actually has a 1-1 column relationship with the underlying FFT model, and use it in addition to the existing peak cache if memory is plentiful. Makes spectrograms appear much faster in many common situations.
3.0-integration
|
| Thu, 05 Jan 2017 11:10:57 +0000 |
Chris Cannam |
Recreate FFT model when something changes, not the next time it's requested (avoids extra work when only retrieving the model)
3.0-integration
|
| Wed, 04 Jan 2017 19:11:10 +0000 |
Chris Cannam |
Merge in subrepo
3.0-integration
|
| Wed, 04 Jan 2017 16:53:06 +0000 |
Chris Cannam |
Pull out record buffer into a separate RT-ish thread
3.0-integration
|
| Wed, 04 Jan 2017 16:03:12 +0000 |
Chris Cannam |
Wire up record monitoring
3.0-integration
|
| Wed, 04 Jan 2017 14:22:39 +0000 |
Chris Cannam |
Make writable model updating explicit rather than essentially an arbitrary hidden accident
3.0-integration
|
| Wed, 04 Jan 2017 13:23:18 +0000 |
Chris Cannam |
Add record update timer (very crude this)
3.0-integration
|
| Wed, 04 Jan 2017 12:05:14 +0000 |
Chris Cannam |
Close wave file writer when recording stops (instead of leaving record file unclosed until the model is deleted)
3.0-integration
|
| Wed, 04 Jan 2017 11:48:03 +0000 |
Chris Cannam |
Fix race condition in first-time recording, where adding the recording wave model would prompt the audio play source to note that its channel count had increased (from 0 to, say, 2) and thus to cause the audio device to be reopened, stopping recording. Fix is to make this only happen if channel count increases beyond that of the device, which shouldn't happen in the recording case
3.0-integration
|
| Wed, 04 Jan 2017 09:57:13 +0000 |
Chris Cannam |
Report audio driver startup error
3.0-integration
|
