A disk image does not have to be all-or-nothing. Open the filesystem inside it, look around, and take out just the files you want.
No spare drive needed. No download of the image. Nothing else fetched.
Why this exists
A share contains a 512 GB image. You want three configuration files from it.
Without this, your options were to download 512 GB, or to write the image to a spare 512 GB drive and then copy three files off it. Both are absurd for three files.
This reads the filesystem inside the image directly over Usenet. It fetches the image's directory structures, shows you the tree, and pulls only the pieces holding the files you pick.
Because content is read in 700 KB pieces, one 5 MiB file out of a 512 MiB image costs a couple of percent of the image. The page tells you what a selection will fetch before you commit to it.
Requirements
The image imported index-only. On the Download page, Import (Index Only). No content is fetched by importing.
A filesystem this build can read. FAT32, exFAT, NTFS and ext4. The page lists what your build supports, and says No filesystem reader installed if none is available.
No elevation, no spare drive, no free space beyond the files you extract. This is the cheap one of the three disk-image features.
Steps
- Import the share with Import (Index Only).
- Open Browse Image in the navigation. (Or click browse inside on the image's own row in the share's file list, which skips the next step.)
- Source: Imported share — then pick the share and the image under Which image?
- Browse the tree that appears. This is the filesystem inside the image, not the share's file list.
- Tick what you want and extract it.
Check the estimate
Before extracting, the page quotes what the selection will fetch. Read it.
Selections behave differently depending on shape:
- A few small files cost very little, even from a huge image.
- Files scattered across the whole image cost more than their combined size, because each one drags in the piece it sits in.
- Most of the image — at that point download the image normally instead. The estimate will make it obvious.
An image already on this machine
The source toggle also offers File on this machine. Point it at a local .img
or .iso and browse inside that, with no Usenet involved at all.
Useful for an image you already downloaded, or one from anywhere else.
Choosing between the three disk-image features
| You want | Use | Costs |
|---|---|---|
| A few files from inside the image | Browse Image (this guide) | A small percentage of the image |
| The image onto a physical USB stick or SSD | Write a disk image to a drive | The whole image, streamed; needs an elevated app and a drive to erase |
| The image as a file on your disk | An ordinary download | The whole image, plus space for it |
| The share's files as a drive letter | Mount a share as a drive | Only what programs read |
Start here when you are not sure. It is the cheapest and the only one that needs nothing but the app.
When it goes wrong
"No filesystem reader installed." This build cannot read any filesystem. On Linux, install the tools for the filesystem you need; the page names what it looked for.
The image opens but the tree is empty or wrong. The filesystem is one of the supported four but an unusual variant, or the image is a whole-disk image with a partition table the reader could not walk. Some proprietary and encrypted container formats also cannot be read.
Extraction is slower than expected. Reading a filesystem's directory structures means many small scattered reads, each a network round trip. Large sequential files are fast; a directory tree with thousands of tiny entries is latency-bound and feels slow. That is the nature of the operation, not a fault.
The estimate is close to the whole image. Your selection is scattered widely. Download the image normally, or write it to a drive.
It is not offered for a file you expected. The file is not recognised as a disk image. Only images with a readable filesystem can be opened.