Currently CryptoFileSystemImpl#readAttributes uses a trial-and-error approach to determine if it needs to read the attributes of a directory or a file.
In some cases one might invoke this method with the cleartext name of a directory (which is derived from an existing directory file) and expects this method to find the corresponding directory. Due to sync conflicts this might not exist, though.
In cases where the directory file exists, but the directory doesn't this method should at least log a warning or even through an exception.
Currently CryptoFileSystemImpl#readAttributes uses a trial-and-error approach to determine if it needs to read the attributes of a directory or a file.
In some cases one might invoke this method with the cleartext name of a directory (which is derived from an existing directory file) and expects this method to find the corresponding directory. Due to sync conflicts this might not exist, though.
In cases where the directory file exists, but the directory doesn't this method should at least log a warning or even through an exception.