Re: HardBound and SoftBound (was "The State of Software")

On Aug 6, 8:28 am, "Wilco Dijkstra"
<Wilco.removethisDijks...@xxxxxxxxxxxx> wrote:
"Terje Mathisen" <"terje.mathisen at tmsw.no"> wrote in messagenews:y_adnVlHbcfkOeTXnZ2dnUVZ8hqdnZ2d@xxxxxxxxxxxxxxx


Not at all: We should instead use some of all those spare cycles/transistors to get rid of 99+% of the current buffer
overflow-style bugs/security risks.

Why waste hardware effort on something that is a essentially a management
issue? Companies are often unwilling to pay the real cost of correct and safe
software.

It's really more of an economic issue and should probably be a public
policy issue. There are lots of incentives to push risk off on the
buyer, and that's what companies do. Ultimately, much of the risk
gets pushed off on the public, along with the unspoken argument that
that's just the way it has to be.

We have plumbing codes, building codes, fire safety codes, codes for
cars, airplanes and buses, and codes for nearly anything that affects
public health and safety... except for anything having to do with
computers.

Incentives have to be built in. If they're not, no one (seller or
buyer) will spend the money. People would rather to buy cheap homes
that might be less safe than to pay to have safer ones, but they are
not given the option. Someone from IBM or DEC might correctly argue
that the microcomputer revolution was really about selling risky
products that they historically would not have, but there is no way to
put the genie back into the bottle. The mentality has put downward
pressure on prices in every market segment, including enterprise
software and hardware--with the possible exception of mainframes,
where the downward pressure is on volume.

One fix or another just isn't going to do it, and neither will Bill
Gates wailing about the competency of US grads. The problems are
systemic and pervasive and will yield only to fixes that are systemic
and pervasive.

I don't think that what Andy has proposed is necessarily a waste of
hardware, transistors, performance, or money, but it may lead to even
more sloppy programming. I don't worry much about syntax because it
costs little to get the compiler to find the errors. I'm not so sure
things would be better if programmers were counting on the hardware to
catch programming and conceptual errors.

Robert.



.

Relevant Pages

  • Re: SP2 - Why bother?
    ... > Your problems are largely caused by your mixing hardware and software ... A SunSparc with Windows emulator? ... plant from a PC only. ... Choices that I make, have associated risk, the risk of programs not working, ...
    (microsoft.public.windowsxp.general)
  • Re: Complete freedom from Microsoft announced today!
    ... > It was a risk I did not think they would take - and I was wrong. ... > With Intel Apple was pushed into a corner by IBM. ... If Dell can pre-load Linux what, besides Apple, would prevent them from pre-loading OS X? ... > suffer from with trying to support so much hardware - likely ...
    (comp.sys.mac.advocacy)
  • Re: Shut down unexpectedly - HELP! Pls
    ... The other possibility is a problem in software, either the operating system, or drivers for your hardware, or a bug in some program that you are running. ... As most experienced computer users know, simply rebooting solves many momentary glitches. ... If your problem is cause by software, the next time it happens, your computer should halt with a BSOD full of cryptic codes. ...
    (microsoft.public.windowsxp.general)
  • Re: What if no more transistors?
    ... error correcting codes in between you, the user, who sees a perfectly ... hardware, which gets the wrong answer with alarming frequency. ... And then you compensate for the "mostly" part with error-correcting ... codes and the like. ...
    (rec.arts.sf.science)
  • Re: Dual Disk SATA
    ... it's due to hardware issues. ... I'm entirely with you regarding backup. ... there is always a risk of backups not ... running applications off of the second SATA disk? ...
    (microsoft.public.windowsxp.general)
Loading