On Jul 1, 2008, at 10:18 AM, Aaron J. Grier wrote:
On Tue, Jul 01, 2008 at 03:05:33PM +0200, Aristotle Pagaltzis wrote:
* Aaron J. Grier <agr...@poofygoof.com> [2008-07-01 03:30]:
On Mon, Jun 30, 2008 at 05:16:19PM -0700, Joshua Juran wrote:
This wouldn't need to be a problem if the image decoder ran in a
separate process with limited privileges.
a (possibly buggy) heirarchy of priviliged process spaces and
message passing doesn't fix sucking software with poor buffer
handling. it just mitigates the bad behavior.
Funny definition of "just." I guess people should just pay attention
and not create bugs, right?
fix them when discovered rather than implement workarounds involving
more software.
Of course bugs should be fixed when they're discovered. The problem
here is preventing damage from myriad untold bugs that haven't been
discovered yet. It turns out that rather than 'just discover them
all and fix them', there's a more efficient kind of solution that
addresses an entire class of bugs at once, without having identified
them individually. Think of it as defense in depth.[1]
I hate that the proposed solution to poor software is usually more
software.
so I take it you love software, and enjoy adding more hate to hate?
I love solving problems.
Josh
[1] http://en.wikipedia.org/wiki/Defense_in_depth_(computing)
