libbz2 bindings for Haskell
#2"Don't use this on pathological input; it may not be secure"
The documentation is not crystal clear on the source of the insecurity of the Codec.Compression.BZip.decompress function. Is it because the underlying C library is insecure in its treatment of pathological input or is it because the bz2 package uses the C library wrongly? Or, is it because of theoretical concerns like the use of unsafeIOToST? Or, is the documentation just trying to express a warning about undiscovered vulns in the underlying C library?
More importantly, as a user I can't tell whether I should be more concerned about getting pwned when using this library than when using the underlying C library directly or the bunzip2 command.
Can this part of the documentation be made clearer? At face value, it sounds like a warning to just not use the decompress function at all because typically users can't tell whether the input is pathological.
- description updated
- description updated
- description updated
- description updated
- description updated
Hi,
Have a look at section 3.2 of the manual: https://sourceware.org/bzip2/manual/manual.html#err-handling
It's designed to work with random corrupt data, but the author says he is not "100% sure" - it's from the underlying bzip2 library
Cheers!