[rohc] IPR... again...

Hi all,

Gosh... guess what..?  It seems that the IPR situation hasn't been
resolved yet...

Ok, well, here's one suggestion: let's ignore context repair (which, it
has to be said, probably counts as novel) and just use the CRC to verify
correct decompression in its most basic form...

You want a reason?  Oh, ok!  Here are some 'quotes' from 'zlib' (you'll
note that this was written in '95/'96 and, who'd've thunk it, contains a
CRC over the uncompressed data...):

	/* gzio.c -- IO on .gz files
	 * Copyright (C) 1995-1996 Jean-loup Gailly.
	 * For conditions of distribution and use, see copyright notice in zlib.h
	 */

<snip>

	typedef struct gz_stream {
	    z_stream stream;
	    int      z_err;   /* error code for last stream operation */
	    int      z_eof;   /* set if end of input file */
	    FILE     *file;   /* .gz file */
	    Byte     *inbuf;  /* input buffer */
	    Byte     *outbuf; /* output buffer */
	    uLong    crc;     /* crc32 of uncompressed data */
	    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
	          This might be relevant...

	    char     *msg;    /* error message */
	    char     *path;   /* path name for debugging only */
	    int      transparent; /* 1 if input file is not a .gz
					file */
	    char     mode;    /* 'w' or 'r' */
	} gz_stream;

<snip>

        if (s->z_err == Z_STREAM_END) {
            /* Check CRC and original size */
            s->crc = crc32(s->crc, start, (uInt)(s->stream.next_out - start));
            start = s->stream.next_out;

            if (getLong(s) != s->crc || getLong(s) != s->stream.total_out)
            {
                s->z_err = Z_DATA_ERROR;
            } else {
                /* Check for concatenated .gz files: */
                check_header(s);
                if (s->z_err == Z_OK) {
                    inflateReset(&(s->stream));
                    s->crc = crc32(0L, Z_NULL, 0);
                }
            }
        }

Now, I know I'm *really* slow on the uptake, but isn't this a CRC over the
uncompressed data to verify decompression..?!

Oh, and before anyone points out the fact: yes, I do know that IPR is
claimed on computing the static / dynamic part of the CRC.  Hmmmm...
sounds like an implementation issue to me.  Ok -- let's try this:
everyone raise your right hand & repeat after me: "I do solemnly swear
that I shall not stoop to any nasty implementation enhancements like
caching the static part of the CRC computation without first consulting my
tame patent attorney."  There, that should fix it..?

Comments..?!

;-/

Mark.

_______________________________________________
Rohc mailing list
Rohc@ietf.org
https://www1.ietf.org/mailman/listinfo/rohc