Description:
------------
When I use dechex() to convert a crc32(), and when I directly use hash("crc32b", ...), I do not get the same output. In particular, the endianness of the hash() output seems inverted.

In other words: the output of crc32() seems to be correct in its integer representation (I compared it with Python implementation), but converting it to hexadecimal (with dechex() or with any calculator) produces a hex string inverted in its endianness respect to the output of hash('crc32b', ...) with the same input.

Reproduce code:
---------------
printf("dechex(crc32(\"AAAAAAAA\")): ".
dechex(crc32("AAAAAAAA")).
"<br><br>");

printf("hash(\"crc32b\",\"AAAAAAAA\"): "
.hash("crc32b", "AAAAAAAA")
."<br>");

Expected result:
----------------
dechex(crc32("AAAAAAAA")): 79b71c0a
hash("crc32b","AAAAAAAA"): 79b71c0a

(or
dechex(crc32("AAAAAAAA")): 0a1cb779
hash("crc32b","AAAAAAAA"): 0a1cb779
I think the first two are the right ones)

Actual result:
--------------
dechex(crc32("AAAAAAAA")): 79b71c0a
hash("crc32b","AAAAAAAA"): 0a1cb779

Please note that the difference is only in the bytes order. "0a-1c-b7-79" is like "79-b7-1c-0a" with inverted endianness, while both lines should output the same values despite different architectures.