The Register® — Biting the hand that feeds IT

Intel erratum-not-bug policy not-good-enough

Yet more Copperwhines about shifting sand

Free whitepaper – Optimizing the data center for cost and efficiency

Letter (A reader has responded to news that there is yet another bug in Intel's Coppermine processor with a letter that claims chip manufacturers should not be allowed to get away with these mistakes.) Though we both know that Intel's erratae represent bugs, the correct Latin to English translation is "mistakes", or "errors". Given that most people have no training in latin these days, (and with many a thesaurus languishing from abandonment) I suggest that you might underline this for some by providing the translation. Erratum=mistakum, or erratum cum errors-in-em The larger issue is of course the ability to determine the fitness of one's own computers. Personally, I think it is remiss on the part of IT journalists not to call for openness on the part of manufacturers in regards to defects. After all if certain assembler instructions cause problems, then 1. We should have a utility to determine if the installed CPU has the bug... er sorry... errata. 1a. If the CPU is problematic, but the BIOS has a remedy installed, report that too. 2. We should have a corresponding utility that analyses our executable files and libraries to determine whether any of the potentially ill-supported machine language instructions are embedded within the code. Neither of these utilities would be hard to write, though their existence could potentially be embarrassing and make it easier to litigate against a CPU manufacturer that does not promptly replace or resolve the defect. No problem if QA improves. Recall any software license that you may have read. In every single one I've read: "The manufacturer does not warrant the fitness of the software for any specific purpose..." ..or something to that effect. Are we now at the stage when we can expect CPU manufacturers to provide a disclaimer with every CPU like: "...the manufacturer does not warrant the proper execution of any assembler instruction, and you agree to limit our liability to the cost of the sand used in making this chip"? I suggest you take the lead and push this. CPU instruction sets need to be better supported. Name and email address supplied

Free whitepaper – Thermal design of Dell PowerEdge server

Don’t Miss

DustbinDirty, dirty PCs: The X-rated picture guide

Ventblockers Horror beyond human imagination

SC09Top 500 supers - rise of the Linux quad-cores

SC09 Jaguar munches Roadrunner

Ubuntu teaser Early adopters bloodied by Ubuntu's Karmic Koala

Smooth Windows upgrade it ain't

Sign up, sign up for The Register IT security newsletter

Narrowcasting for the email classes