[Prev][Next][Index][Thread]
Re: You don't want to read this....
On Sat, 29 Apr 2000, Alan Grimes wrote:
>But I'm gonna post it anyway. =P
>
>This project is obsolete.
><snip>
>All who have studied the
>contemporary solutions to this issue are aware of the many rather inelegant
>and arbitary hacks that have been implemented and the high cost in terms of
>latancy that are brought on with contemporary techniques.
><more snippage>
>Their project seeks to eliminate the kernel and all the enforcement
>mechanisms that sap performance. The method they are using to accomplish
>this is to first prohibit direct access to the machine. All programs must
>be compiled or interprited.
Not to be a killjoy, but earn't hardware enforced protection mechanisms
going to be faster than an interpreted system! And even if the programs
are compiled, I assume they will only be compiled at runtime, as to
compile beforehand would allow code to be built to circumvent the software
protection systems.
Hardware protection also gives protection against bugs, as well as
malicious or promiscuous code. Without that hardware based pointer
validation, that stray NULL pointer will not be caught.
Christian
PS. No, I have not yet read the tunes faq or other docs.
PPS. This has already been done. It's called OS/400;)
References: