[MPlayer-cvslog] r22958 - trunk/stream/realrtsp/real.c

[Reimar Döffinger]

Hello,
On Mon, Apr 09, 2007 at 10:32:47PM +0200, Roberto Togni wrote:
> On Mon, 9 Apr 2007 22:01:51 +0200
> Reimar Döffinger <Reimar.Doeffinger at stud.uni-karlsruhe.de> wrote:
> > On Mon, Apr 09, 2007 at 09:30:33PM +0200, rtogni wrote:
> > >      /* xor challenge bytewise with xor_table */
> > > -    for (i=0; i<table_len; i++)
> > > +    for (i=0; i<XOR_TABLE_SIZE; i++)
> > >        ptr[i] = ptr[i] ^ xor_table[i];
> > 
> >  ptr[i] ^= xor_table[i];
> > 
> > would be simpler, too, if you care ;-)
> 
> Yes, that (and others) are planned for later cleanup patch.

Ok, nice :-)

> > I guess speed is not relevant, but if it was the xor also could be done
> > on uint32_t instead of bytewise (and bytewise doesn't have endian issues)...
> > 
> 
> It looks simpler this way, and no endianess or alignment problems.

Yes, when speed is not relevant it's a good choice at is it.
Just mentioned it as during the aes implementation I noticed how extreme
the speed difference is between doing it byte- and int-wise.

Greetings,
Reimar Döffinger