Home | History | Annotate | Download | only in srtp
      1 TODO List
      2 
      3 1.4.1
      4 
      5   - document which fields are in NBO/HBO, and check for consistency.
      6 
      7   - move HAVE_U_LONG_LONG inside of datatypes.c, or some other
      8     separate file
      9 
     10   - re-write configure.in to make cross-compilation easier
     11 
     12   - eliminate GENERIC_AESICM by generalizing the code a bit
     13 
     14 Older comments
     15 
     16   - add tests for key_limit_t datatype 
     17 
     18   - move octet_get_weight() from datatypes.c to math.c (any other
     19     funcs?)
     20 
     21 Changes and additions planned
     22 
     23   Make cipher and auth dealloc() functions zeroize the key-storage
     24   areas before calling free().
     25 
     26   Eliminate key_len from auth_init()
     27 
     28   Doucument internal APIs (cipher, auth, srtp_protect, ...)
     29 
     30 
     31 SRTP options not (yet) included in this libaray:
     32 
     33  - the aes-f8-mode cipher
     34  - the Master Key Index
     35  - re-keying using the key derivation function (only the initial
     36    use of the PRF has been implemented, as it's sufficient
     37    for most uses)
     38 
     39 
     40 (OLD) PLANNED CHANGES
     41 
     42    strip out test/lfsr.c
     43 
     44    Write new documentation!!!
     45 
     46    Fix the x86 assembly code in aes.c.
     47 
     48    Eliminate /* DAM */ - there's one in srtp.c
     49 
     50    Change debugging so that it can print more than one line.  Or perhaps
     51    just change it so that a single check of the debug-enabled flag is
     52    needed.
     53 
     54    Improve interface between cipher and rdbx - perhaps generalize rdbx
     55    into 'nonce' datatype.
     56 
     57    Make rijndael_icm accept variable sized keys.
     58 
     59    Add rdbx functions that allow different-sized explicit sequence
     60    numbers to be used.
     61 
     62    Write uniform byte-buffering code for PRFs, preferably as macros.
     63 
     64    Consider eliminating low-level alloc functions in favor of len()
     65    functions, so that there need not be multiple allocations within a
     66    particular alloc() function.
     67