home *** CD-ROM | disk | FTP | other *** search
- Submitted-by: karish@pangea.Stanford.EDU (Chuck Karish)
-
- In article <1m3algINNelm@ftp.UU.NET> clive@x.co.uk (Clive Feather) writes:
- >This behaviour is in cpio so that it is possible to extract any file
- >without special handling. Suppose you create an archive containing three
- >links:
- > mkdir newdir
- > cd newdir
- > cp /some/random/file a
- > ln a b
- > ln a c
- > ls ? | cpio -oc >archive
- >Then it is possible to extract the file under any of its names [ ... ]
- >If you handle links specially in the archive, then the first of these
- >works, but the other two don't, unless cpio can back up through standard
- >input, or saves all files with >1 links just in case another name
- >matches the pattern.
-
-
- The SVR4 070701 implementation solves this problem by
- storing the data under the last name for the link.
-
- This means that cpio must cache the mnames of at least the files
- with multiple links, and create a link index before writing
- the archive.
-
- I hope this problem is handled properly in the next new archive
- format. And I hope that format is designed well enough that
- it isn't immediately obsolete, as POSIX CPIO was.
-
- --
-
- Chuck Karish karish@mindcraft.com
- (415) 323-9000 x117 karish@pangea.stanford.edu
-
-
- Volume-Number: Volume 30, Number 68
-
-