Scheduler now has the ability to run jobs on multiple repositories concurrently and run multiple jobs on one repo (by queuing them).
For each repository, there is one queue. I have represented a queue by a 'site'. Between sites (ie repository), tasks run concurrently. On one site, tasks run one by one. The user also run tasks by adding them to the queue but he can't run multiple backups because start backup button is disabled when a job is running.
* fix issue #940 - KeyError in get_dict_from_list
- cause of error : defaultdict added more defaultdict while attempting to traverse the tree, but once a 'potential leaf' node was added, it was added as a dict, not a defaultdict.
- two possible solutions:
- 1 - change everywhere that adds a 'potential' leaf node to add a defaultdict (ie nested_dict()) - this occurs in several places, but not many.
- 2 - change get_dict_from_list to add a default dict (not defaultdict) when traversing the tree, for the case where a multi-level node is added on top of an existing node. This requires only changing a single location, and means that the dictionaries returned by accessing the tree will behave like normal dict (ie, won't by default add missing keys).
* Add test case for issues #940 and #925
* add support for 'diff --json-lines' available in borg 1.1.16.
* fixes issue #901 (diff not handling CR/LF in filenames)
* improve `pretty_bytes()` formatting function
