Re: libnettle/libhogweed WIP

[Ted Zlatanov]

On Mon, 17 Apr 2017 12:55:32 -0400 Ted Zlatanov <address@hidden> wrote: 

TZ> I see the confusion. Yes, I want to allow multibyte input. Users
TZ> shouldn't have to jump through hoops to use these functions.

OK, the code has been changed as follows:

* since I was in fns.c, I added `secure-hash-algorithms' so we know what
  algorithms are supported by `secure-hash'. At least for tests it's
  nice to have this.

* factored out extract_data_from_object() from secure_hash() which
  incidentally defines a data format for data extraction
  (BUFFER-OR-STRING INPUT-START INPUT-END CODING-SYSTEM NOERROR) because
  I really didn't want my functions to take 20+ parameters.

* secure_hash() could also take that format but I left it alone for now.

* migrated secure_hash() and all the new gnutls.c crypto code to
  use extract_data_from_object()

* added tests to verify that buffer extraction (whole buffer) works

* IV, AUTH, and KEY parameters are still wiped if they were strings
  originally.

phew...

Ted