Commit graph

6 commits

Author SHA1 Message Date
joerg
647e9d137e digest-20080509:
The tiger hash needs to run at least once through the update function,
even for empty input. The fix is different from the patch in PR 36565.
2008-05-09 15:00:32 +00:00
joerg
f60512c785 Fully ANSIfy and use size_t and uint32_t in places where u_int was used
before. This fixes the build on QNX, where u_int is not exposed by the
current set of headers. Make the prototypes of the crypto functions
consistent.
2007-09-21 18:44:36 +00:00
joerg
315daaa372 As reported by tnn@, the old approach to find the byte order
doesn't work on HP-UX/Itanium. Use the standard approach using
AC_C_BIGENDIAN. Tested by myself on i386, he@ on SPARC and tnn@ on
HP-UX.
2007-07-27 17:08:27 +00:00
joerg
b1a42ef8c2 Make pkgtools/digest cross-compilation ready and actually improve
portability somewhat:
- consistently use POSIX uint{8,16,32,64}_t throughout the code.
- asssume for now that the compiler supports ULL as suffix
- use AC_TYPE_UINT{8,16,32,64}_T instead of bits.c to compute
  possible replacement types.

Bump to 20070703. Tested by Thorsten Glaser on Interix and tnn@ on
Solaris.
2007-07-03 18:54:02 +00:00
joerg
49de1e68ab Revert libncompat usage and kill the use of snprintf in whirlpool and
tiger instead. Also kill bogus NULL pointer handling in both algorithms:
passing null pointers as buffers (when len != 0) or context is a serious
programming error and should result in a segfault. Bump version to
20060826. OK agc@
2006-08-29 13:43:11 +00:00
agc
0efe6ebb73 Add support for the tiger algorithm to digest.
Bump version to 20050310.

The cabal known as the cryptographers might like to have a look at the
first two results in test 6 of the regression tests.
2005-03-22 10:16:15 +00:00