TAR archive file format
- Typical file name extensions
.tar- Magic bytes
- None.
- However, identification of tar files is not limited to looking at the file extension.
Each tar archive entry header also stores a checksum on itself.
Thus, reading the first 512 bytes of a potential tar file, creating that checksum and
comparing it to the stored checksum will tell if the file really is in tar format.
- MIME types
application/x-tar- File structure
- A tar archive interchangeably stores headers and file data.
- Headers are 512 bytes large, followed by zero or more blocks of 512 bytes size each (containing uncompressed file data).
If an archived file has a size in bytes which is not divideable by 512 without rest, as many bytes as missing to the next 512 byte boundary are added to fill up that last block.
A file that is 513 bytes large thus occupies 512 bytes for the header, 513 bytes for the file's data, and 511 additional bytes so that the file content will be a multiple of
512 bytes (in this case, 1024 bytes).
Tar was designed for tape devices which can be accessed only a block at a time.
512 bytes (half a kilobyte) was the standard size of a single block.
- Compression types
- None.
- History
- Tar (for tape archiver) was designed to create backups on tapes with Unix systems.
- Popularity
- Very high on Unix systems.
"Tar balls" (a number of files stored in a single tar archive) are often used to distribute everything from simple utilities to the Linux kernel.
- Often used in combination with gzip
(and more recently bzip2) to reduce the size of the archives.
- Meta data
- The fixed-size (512 bytes) entry headers store name, size, date of last modification, access rights and more.
- Limitations
- No compression supported.
- No support for extended character sets in file names.
- Entry names can only be 100 characters long (including path).
- Data recovery
- Checksums are created only on the headers, so erroneous data will not be detected.
- Encryption
- Not supported.
- Solid archives possible?
- No. Tar does not support compression, so solid archiving (compressing more than one archive entry in a row) does not apply in the context of tar archives.
- Support for multiple volumes?
- No.
- Libraries
- GNU tar (implementation of the tar utility in C).
- Java tar
- Specification
- tar(5) man page on Unix systems (Google finds those as well, try tar 5 man).