libdcp/test, branch gh-21-open DCP handling library https://git.carlh.net/cgit/libdcp/atom?h=gh-21-open 2023-09-13T19:37:15Z Remove odd assertion stopping TZ minutes being more than 30 or less than -30 (DoM/GH #20). 2023-09-13T19:37:15Z Carl Hetherington cth@carlh.net 2023-09-13T19:37:15Z urn:sha1:d0c41db8b5c3b1548faf6e11562c59a6161e16c6 Check for missing font files referred from Interop closed captions (as well as subs). 2023-08-24T21:19:44Z Carl Hetherington cth@carlh.net 2023-08-24T21:19:44Z urn:sha1:8305f28ea22a834cc4a9e4b324110b5ff4ba1bd1 Accept any case for subtitle alignment specifiers (#2602). 2023-08-22T21:45:14Z Carl Hetherington cth@carlh.net 2023-08-22T21:45:13Z urn:sha1:5980cee25f537c54fc1155de5492bddd92bfa338 SMPTE seems pretty clear to me that these should be lower case, but counterexamples have been seen in the wild and it seems to make sense to allow them. Perhaps the verifier should raise an error with non-lower-case examples. Add some tests. 2023-08-22T21:35:32Z Carl Hetherington cth@carlh.net 2023-08-22T21:35:32Z urn:sha1:ac449700a47c164bf09ee2b0c18486ed255ca20f Fix dcp::LocalTime constructor to cope with longer fractional second parts (DoM #2597). 2023-08-15T10:28:21Z Carl Hetherington cth@carlh.net 2023-08-15T10:28:21Z urn:sha1:545e3f54b104364b316318876a3d4e515de758b1 Add check for empty <LabelText> in <ContentVersion> 2023-08-11T18:51:23Z Carl Hetherington cth@carlh.net 2023-08-11T18:51:21Z urn:sha1:4dca630057509164494b65c2deeb748a51928c73 We have a report of Deluxe failing a DCP because of this. Fix boost bind placeholder warnings. 2023-08-08T12:38:47Z Carl Hetherington cth@carlh.net 2023-08-08T12:38:47Z urn:sha1:20b6a9862bbde42e9ce038b33ec049194a8146a5 Include trimming. 2023-07-25T13:38:50Z Carl Hetherington cth@carlh.net 2023-07-25T13:38:50Z urn:sha1:d4270052ecfd3f75314fc8ea0e22c20384ff8b3c Cleanup: move EqualityOptions into its own file. 2023-07-25T13:28:11Z Carl Hetherington cth@carlh.net 2023-07-25T13:28:11Z urn:sha1:d21b1e6f9cbf10148b38e1bc3ca3d4920e39a432 When reading a DCP set up asset hashes from the CPL/PKL, not by digesting the actual asset. 2023-06-28T23:25:40Z Carl Hetherington cth@carlh.net 2023-06-26T22:11:43Z urn:sha1:e702623781c8d5853b79a29ca8c5f495d0ade3d1 Previously when reading a DCP we would recalculate asset hashes. This meant that corrupted DCPs could be loaded in and their hashes re-written (if nobody checked the new hashes against the ones in the PKL/CPL). It seems better to take the hashes from the metadata. Then if the assets are corrupted and the DCP is re-written the PKL/CPL hashes will be preserved, showing rather than hiding the corruption.