Re: On2's VP5

Ian Farquhar <[email protected]> Thu, 12 Aug 2004 16:45:17 +1000
Newsgroups gmane.comp.video.xine.codec.devel
Message-ID <[email protected]>
At 11:29 AM 12/08/2004, Mike Melanson wrote:
> > You guess wrong.  Microsoft's port of DX/VFW to Solaris is pretty poor,
> > which is to say, non-existent.
>
>         I.e., this multimedia wrapper does even less than speculated...

If it can be a wrapper, it will be (to take advantage of hardware 
acceleration, for example).  If not, the commoner codecs are coded in pure 
Java so they run pretty much anywhere.  That's the idea of JMF.

> > Despite MS's propaganda, it's quite possible to run an MS-free
> > infrastructure, and Sun does.  The vast majority of servers are Solaris
>[...]
>Not sure how Sun cheerleading fits into this list...

It doesn't, and I'm actually quite critical of Sun for other reasons.  My 
point was that assuming all code runs over MS OS's - as is evidenced by the 
comment about redirecting to DirectMedia - was ill-informed.

 >       As Michael N. pointed out, support is probably limited to the
 > stuff that is "open" in the first place (for which we already have open
 > source support).

They licensed quite a few things, but most of the closed ones have been 
RE'd (eg. Cinepac).

Sadly, iv31 and iv32 _aren't_ in the native list.  From memory, they're 
still outstanding.  I'm pretty sure we know all the rest from the list I 
have here.

 >       Still, from a cursory review, this Java obfuscation is *nothing*
 > compared to reverse engineering C/C++ code compiled & optimized into
 > machine opcodes. Seriously, RE'ing this obfuscated Java reminds me of
 > solving a substitution-cipher cryptogram. All of the algorithmic logic
 > appears to be right there in the code. The only trick to understanding the
 > algorithm is to recover the identifier strings.

With respect, I think you need to do a slightly longer look at the Java 
obfuscation field.  There are quite a few things which can be done to make 
the resultant bytecode very hard to reverse, especially if you're willing 
to take a 10-20% performance hit in the process.  I do concur, though, that 
there are aspects of Java bytecode's design which assist the reverse 
engineer when compared to native machine code.

Ian. 




-------------------------------------------------------
SF.Net email is sponsored by Shop4tech.com-Lowest price on Blank Media
100pk Sonic DVD-R 4x for only $29 -100pk Sonic DVD+R for only $33
Save 50% off Retail on Ink & Toner - Free Shipping and Free Gift.
http://www.shop4tech.com/z/Inkjet_Cartridges/9_108_r285