Skip to content

Levelized Files #154

Description

@ssoelvsten

Motivation

  1. A lot of our files are of the temporary read-once-and-produce-output type (specifically arc_file and node_file with move-semantics). Hence, to be able to squeeze more BDDs / ZDDs into a smaller disk, we should continually deallocate the input file.
  2. There is an upper limit to the file size for each file system, e.g. ext4 does not support files larger than 16 TiB.

Solution

The meta_file is turned into a list shared_ptr to files (or a file of files?). When a stream hooks into a read-once file, it removes the file for each level, when it has finished reading it.

Metadata

Metadata

Assignees

No one assigned

    Labels

    questionFurther information is requested✨ optimisationIt's all about speed / space📁 internalThis is where the ✨magic✨happens

    Projects

    No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions