| Age | Commit message (Collapse) | Author | Files | Lines | 
|---|
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | The experimental cloud support was triggered on each run of any command
which caused some to break on devices without iCloud setup.
We now trigger the cloud support in a separate command to fix this and allow
further investigation into a proper implementation. | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | Allows to specify a password for the Unback request. It is also sent in the
main message, not in an options dictionary. | 
|  | mobilebackup2_send_status_response(). | 
|  | to mobilebackup2_send_status_response(). | 
|  |  | 
|  | returned from plist_get_string_val(). | 
|  | it can’t be retrieved from the lockdown connection. That mimics the leaking implementation, but maybe it should be an error. | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | crashes | 
|  |  | 
|  |  | 
|  | This is a major change which breaks API but is required in order to support
SSL communication for services as used by network connections. | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  |  | 
|  | If the device is sending files to the host, it sometimes requires a bit more
time to process them before sending. This appeared to happen mostly for larger
sqlite databases which appear to get some preprocessing on the device.
In such a "wait" situation, we receive no data and need to retry reading the
filename length again. Due to a code bug though which didn't reset the last
read length to zero, this length was incorrectly alternating between 1 and
16777216 due to the byte swapping. This ulitmativly lead to a broken backup
process. Now we properly wait for the device to preprocess any file before
sending the filename to the host. | 
|  |  | 
|  |  |