gedit complains about invalid characters when opening xbmc.log file produced by pi Since the change from perforce revision number to git hash, the version string has increased in size. Fortunately we don't overrun the "response" buffer. Unfortunately when buffer doesn't fit, it is not null terminated, so logging message pulls in garbage. Version string looks like: Jun 17 2013 20:49:11 Copyright (c) 2012 Broadcom version d380dde43fe729f043befb5cf775f99e54586cde (clean) (release) So, 80 characters is not enough (it is about 115 plus newlines). 160 seems more than enough, but truncate the string just in case.
… mailing list) - this was signed off by elupus
…lScreen into Application
…t qualifiers. Before, the MIME type would only be resolved if GetMimeType(true) was called, which is potentially blocking. Now MIME types will be resolved from the file's extension without possibly blocking the thread. See doxy of CFileItem::FillInMimeType() for more info.
I think this doesn't tend to happen on Pi now, as AV_SAMPLE_FMT_S16 and AV_SAMPLE_FMT_FLTP bypass the resampler, but best to fix it if some obscure codec returns something else. See #2873
As title say faster way to correct the filetype field behavior. (A lot since std::set use key as index while CFileItemList.Get on filename needs locks and tons of overhead, not talking about the size of parameters compared to a large directory).
… functions. it will fall back to other methods if not present and we have no guarantee that the libavcodec version we build against will be the same we run with (except it has the same ABI but ff_vdpau symbols are not part of the ABI).
There is no point in doing this and opening an already opened context leaks memory in lavr.
If video does not start on an I frame, you can get a significant difference in pts and dts values for the first frame. I've got two videos where they differ by more than 0.1s. In this case, we end up setting pts=0 and leaving dts alone (which may be minutes away). This gets through to GPU on Pi, and it tends to stall waiting for this difference to elapse. I'm not sure what the largest difference can be, but 0.5s feels completely safe. I'm not sure of any problems increasing this will cause.