#1 Inconsistent comment and code in Canon ReadODD

[Ihor Dutchak]

Code, that parse canon original decision data version 3 (lib/Image/ExifTool/Canon.pm#l8513 in current master) has some comment, explaining it:

elsif ($version == 3) { # newer models (little endian)
            # this data is structured as follows:
            #  4 bytes: all 0xff
            #  4 bytes: version number (=3)
            # 24 bytes: sha1 A length (=20) + sha1 A
            # 24 bytes: sha1 B length (=20) + sha1 B
            #  4 bytes: length of remaining data (including this length word!)
            # ... some more

but it is not what the code does:

            for ($i=0; ; ++$i) {
                $i == 3 and undef $err, last; # success!
                $raf->Read($buf2, 4) == 4 or last;
                $buff .= $buf2;
                my $len = Get32u(\$buf2, 0);
                # (the data length includes the length word itself - doh!)
                $len -= 4 if $i == 2 and $len >= 4;
                # make sure records are a reasonable size (<= 64kB)
                $len <= 0x10000 and $raf->Read($buf2, $len) == $len or last;
                $buff .= $buf2;
            }

From what I see in the code - length of remaining data appear before sha1 A/B, but the comment says - after.

Should it be consistent?

[Ihor Dutchak]

I guess I got confused only by code style.

Yes, It is consistent.

Please close the ticket as invalid, and excuse me for interrupting.

[Phil Harvey]

Right. Glad you sorted this out. Thanks for caring enough to post in the first place.