[sldev] Cache politics: performance vs obfuscation

[Darien Caldwell]

Argent wins the thread. Most are either completely downplaying any sort of 
deterrent to theft and actually encouraging it, while the other side is 
proposing ridiculously complex encryption specifications which would 
severely harm the ability of SL to function.

Argent's proposal is balanced enough to make everyone happy. Sure, anyone 
with can get past a "Hello World" program will be able to break it. But the 
point is, 90% of SL users can't get that far. So as a deterrent, it will 
prevent casual misappropriation of restricted materials.  On the other end, 
it won't severely hamper anyone from doing whatever 'ubercool' stuff they 
want to do.

It gets my vote.

----- Original Message ----- 
From: "Argent Stonecutter" <secret.argent at gmail.com>
To: "Laurent Laborde" <kerdezixe at gmail.com>
Cc: "Second Life Developer Mailing List" <sldev at lists.secondlife.com>
Sent: Tuesday, June 10, 2008 8:22 PM
Subject: Re: [sldev] Cache politics: performance vs obfuscation


> On 2008-06-10, at 13:57, Laurent Laborde wrote:
>> If LL develop a custom image format, it will cost a lot of ressource
>> to work on optimization, etc ...
>
> The 'custom image format' under discussion is just "dump the decoded 
> texture in whatever format it exists in memory, along with associated 
> metadata, with a constant-cost obfuscation layer such as XORing the 
> content with some constant secret."
>
> There is no optimization necessary, because there's no codec, and  miminal 
> overhead: it's a straight copy and a tight "XOR" loop. Even  simple RLL 
> encoding is more overhead.
>
> _______________________________________________
> Click here to unsubscribe or manage your list subscription:
> /index.html