Official: Microsoft 'bribes' companies to use Live Search
Moderator: Thanas
whelp, shows what I know. I still maintain that the only smart ship I've seen (The Shiloh...or the Antietam? All the CW-themed CGs sound the same to me) had a UNIX-type OS in their CCS.
"The rifle itself has no moral stature, since it has no will of its own. Naturally, it may be used by evil men for evil purposes, but there are more good men than evil, and while the latter cannot be persuaded to the path of righteousness by propaganda, they can certainly be corrected by good men with rifles."
That seems exactly like something a non-tech journalist would write about computers when he wanted an sensationalist article. The tech issues are grossly distorted.
The divide-by-zero issue for example - yes, its about as basic as you can get with computer problems, however "If you understand computers, you know that a computer normally is immune to the character of the data it processes, he wrote in the June U.S. Naval Institutes Proceedings Magazine" is hilariously wrong. There's the mentioned divide-by-zero problem, overflow problems (to use his calculator example, try entering the maximum number your calculator supports and then adding one - depending on the software, it will either crash, display an error or wrap around to the lowest number possible) data type incompatibility problems, and so on.
Computers require data to be defined very specifically, and while more modern development environments have some integrated tools to help detect such problems before they show up in a production program, back in '97 this was far less available. Yes, the issue is something that should have been caught in testing at the latest, however acting like it is some sort of introduced problem is lying, if spoken by someone with even rudimentary knowledge of computer programming. If you give the computer a command to divide by zero it will stupidly comply, and unless you handle the resulting problem (namely, that it is an illegal action) the application (or the OS in the case of older ones) will crash. So while they should have protected the system from such a crash (it is a rather basic thing to protect against), implying that it shouldn't happen by default by invoking the calculator analogy is simply wrong.
And when you're running something that important as an warships OS, you probably have all the niceties that help ripped out and are programming in C/C++, which, depending on the compiler, give you all the rope you need to hang yourself, if you aren't thinking of the possibilities, which in turn is also the reason it is so powerful.
While we're obviously talking about an insufficiently tested system going by what occurred, Unix would suffer similar problems with badly written applications (of course, if Unix was previously deployed with applications that were proven to function then, yes, they should have stayed with that until such basic issues were worked out of the new systems). NT is not an inherently unstable OS, but it is only an OS and if applications written for it suck and can easily crash (and a divide-by-zero issue is a damning indictment of the ability of the programmers if they get it from a data entry field in something as mission critical as a warship OS) that is an application quality control problem, not the underlying OSs. I've seen nothing in the article indication a failure on the part of the OS and not the applications used on said OS (the article does not make the distinction when it should).
Using the OS names for the entire system is simply wrong. They should be talking about some sort of system names, the combination of applications and the underlying OS, and not the OSs themselves which, as long as they are stable, which both NT and Unix are, provide little more then somewhat different foundations (APIs) on which to build apps that do the actual control in a scenario such as this. Using NT in this context is simply populist scaremongering by tying the perceived instability of consumer Windows (mostly based on the 9x line which has nothing to do with NT based OSs) compared to Unix to systems that simply do not use those OSs in anything resembling the same way nor form (the NT used for the Smart Ship project is undoubtedly stripped down to essentials far beyond what is possible with consumer versions - nobody is going to be playing Solitare, reading e-mails or watching porn on that bad boy).
In fact, while Redman comes of as a reluctant professional (all of his observations are correct if taking into account the sensationalist writer - I'd bet that if you replaced "NT" with "NT based system" you would get a more accurate account of what was actually said), DiGiorgio is a lying scumbag (basically all of his observations have problems and are either partially or fully wrong to someone with any programming experience) with a martyr complex ("self-described whistle-blower"). Or he's being intentionally misrepresented by the writer who's ineptly punching up the problems mentioned by DiGiorgio to get a more sensationalist article. The only part of the article attributed to him I can agree with is:
The divide-by-zero issue for example - yes, its about as basic as you can get with computer problems, however "If you understand computers, you know that a computer normally is immune to the character of the data it processes, he wrote in the June U.S. Naval Institutes Proceedings Magazine" is hilariously wrong. There's the mentioned divide-by-zero problem, overflow problems (to use his calculator example, try entering the maximum number your calculator supports and then adding one - depending on the software, it will either crash, display an error or wrap around to the lowest number possible) data type incompatibility problems, and so on.
Computers require data to be defined very specifically, and while more modern development environments have some integrated tools to help detect such problems before they show up in a production program, back in '97 this was far less available. Yes, the issue is something that should have been caught in testing at the latest, however acting like it is some sort of introduced problem is lying, if spoken by someone with even rudimentary knowledge of computer programming. If you give the computer a command to divide by zero it will stupidly comply, and unless you handle the resulting problem (namely, that it is an illegal action) the application (or the OS in the case of older ones) will crash. So while they should have protected the system from such a crash (it is a rather basic thing to protect against), implying that it shouldn't happen by default by invoking the calculator analogy is simply wrong.
And when you're running something that important as an warships OS, you probably have all the niceties that help ripped out and are programming in C/C++, which, depending on the compiler, give you all the rope you need to hang yourself, if you aren't thinking of the possibilities, which in turn is also the reason it is so powerful.
While we're obviously talking about an insufficiently tested system going by what occurred, Unix would suffer similar problems with badly written applications (of course, if Unix was previously deployed with applications that were proven to function then, yes, they should have stayed with that until such basic issues were worked out of the new systems). NT is not an inherently unstable OS, but it is only an OS and if applications written for it suck and can easily crash (and a divide-by-zero issue is a damning indictment of the ability of the programmers if they get it from a data entry field in something as mission critical as a warship OS) that is an application quality control problem, not the underlying OSs. I've seen nothing in the article indication a failure on the part of the OS and not the applications used on said OS (the article does not make the distinction when it should).
Using the OS names for the entire system is simply wrong. They should be talking about some sort of system names, the combination of applications and the underlying OS, and not the OSs themselves which, as long as they are stable, which both NT and Unix are, provide little more then somewhat different foundations (APIs) on which to build apps that do the actual control in a scenario such as this. Using NT in this context is simply populist scaremongering by tying the perceived instability of consumer Windows (mostly based on the 9x line which has nothing to do with NT based OSs) compared to Unix to systems that simply do not use those OSs in anything resembling the same way nor form (the NT used for the Smart Ship project is undoubtedly stripped down to essentials far beyond what is possible with consumer versions - nobody is going to be playing Solitare, reading e-mails or watching porn on that bad boy).
In fact, while Redman comes of as a reluctant professional (all of his observations are correct if taking into account the sensationalist writer - I'd bet that if you replaced "NT" with "NT based system" you would get a more accurate account of what was actually said), DiGiorgio is a lying scumbag (basically all of his observations have problems and are either partially or fully wrong to someone with any programming experience) with a martyr complex ("self-described whistle-blower"). Or he's being intentionally misrepresented by the writer who's ineptly punching up the problems mentioned by DiGiorgio to get a more sensationalist article. The only part of the article attributed to him I can agree with is:
Installing something as untested and unstable as the mentioned NT based system seems to be, with such basic issues, should have been grounds for shitcanning some overseeing people for failing at basic quality control, hauling the system back in for patching, and reinstalling it only after it passed rigorous testing in a simulated environment (and any database problems should damn well be caught by such testing for something so important).But DiGiorgio said that the Smart Ship project needs to do more
engineering up front.
Installing a control system on a warship and resolving problems as the
project progresses is a costly and naive process, DiGiorgio wrote in the
Proceedings article.
-
Uraniun235
- Emperor's Hand
- Posts: 13772
- Joined: 2002-09-12 12:47am
- Location: OREGON
- Contact:
I'd be interested to know what sort of drivers are required for the computers driving those Smart Ships. Badly-written software shouldn't necessarily take down an OS, but a shitty driver can ruin just about anyone's day.
"There is no "taboo" on using nuclear weapons." -Julhelm
What is Project Zohar?
"On a serious note (well not really) I did sometimes jump in and rate nBSG episodes a '5' before the episode even aired or I saw it." - RogueIce explaining that episode ratings on SDN tv show threads are bunk
"On a serious note (well not really) I did sometimes jump in and rate nBSG episodes a '5' before the episode even aired or I saw it." - RogueIce explaining that episode ratings on SDN tv show threads are bunk
Ultimately, most of the "smart ship" personnel reduction comes not from the combat systems and operations departments (you're still going to need techs to do regular PMS of the systems) but the engineering departments. Most of the software associated with large vessels engineering spaces has been in the commercial shipping market for years and years, and probably wouldn't require re-inventing the wheel.
"The rifle itself has no moral stature, since it has no will of its own. Naturally, it may be used by evil men for evil purposes, but there are more good men than evil, and while the latter cannot be persuaded to the path of righteousness by propaganda, they can certainly be corrected by good men with rifles."