On 4/20/2011 4:01 PM, Patrick Elliott wrote:
> There is a reason why a lot of biologists use the term IDiocy to
> describe this. Sadly, a few too many comp-sci people fail to grasp why
> its similarly intractable in terms of computer code (where, the
> equivalent in that would be taking a running application, and inserting
> new code into it, without breaking links, jump points, functionality, or
> mangling the data being processed, mid-change).
To clarify, other requirements - The application be monolithic, not 
include scripting/JIT compilation, and have massive crosslinking, such 
that a minor change "might" be OK in the running code, but if you do 
need to reboot (split off a clone, or produce other sorts of offspring), 
some form of reboot "will" happen, and your change has just broken 500 
other steps in the process, since its a critical part of some library, 
which just happened to not cause problems in the "running" 
implementation, but causes a whole shitload of errors, during start up 
(like a zip library, which works fine for a set sized data sample, but 
breaks when you try to unzip entire branches of the boot code, which are 
larger, or smaller, than being handled in the "running" instance).

In any case, even if you could produce such a thing, and have it self 
modify safely, it only helps the front load argument, not the "it didn't 
evolve" one. You still have to show where the front load code is, which 
means doing research, and... these people don't *do* research.

Post a reply to this message