If someone in Linus's position were litigious
(he's not) and vindictive (he's not), they might go after the SCO principals as individuals. SCO may be dying, but you can bet that its principals will salt some money away before the company's legs stop twitching.
Personal bankruptcy is not an easy out... so I've heard.
I heard a radio interview with the baloon's developer on the BBC World Service last night. When the ballon got loose, it's battery got ripped off and hence its "brain" is "dead". (The developer made this very clear... in response to a series of decidedly lame questions by the interviewer.) Any discussion of the "intelligence" (or otherwise) of this this particular balloon is moot.
We can filter by IPs or keywords or addresses or whatever, but they one thing they can never disguise is their message: it has to be available or they're just sending static. Part of that message has to be some way to contact the company, or else there's no way for you to order their penis enlargement cream or online marketing guide.
Unfortunately, some spammers are using so called "reverse proxies" installed on hacked machines to
host the webpages / email boxes mentioned in the
crap they send you. To find the true address for
the spammer you need to locate and reverse engineer the hacked machine.
Halon is a liquefied, compressed gas that stops the spread of fire by chemically disrupting combustion. Halon 1211 (a liquid streaming agent) and Halon 1301 (a gaseous flooding agent) leave no residue and are remarkably safe for human exposure. Halon is most effective for flammable liquids and electrical fires (rated B:C) and is electrically non-conductive.
Actually, this is common sense. If you wanted to damp combustion by excluding oxygen, it would be cheaper (and more environmentally friendly) to use an inert gas like nitrogen or carbon dioxide.
There is no doubt that Halon does
replace oxygen to some degree and therefore
does present a potential danger of asphyxia. However, there is another problem with Halons. When they come into contact with a fire, they breaks down, releasing breakdown products that are extremely dangerous, even at low concentrations.
If I remember correctly (from my 4.2bsd days),
you can also run into trouble if your hard drives
have bad blocks, and you revector them in the
OS's file system driver. Doing a 'dd' on a raw
partition would run into the bad blocks, and bad
things may happen to your backup.
Assuming that the Government wants to retain control of their IP while allowing another contractor to do further work, they could supply their modifications as patches against a specified source distro. These patches would consist solely of Government owned IP, and would not be covered by the GPL. Hence the Government could place any restrictions on the patches.
Alternatively, (IANAL) the contractors might legally be construed as agents of the Government for the purposes of IP control wrt the contract. Hence, providing them with a copy of the modified software (for the purposes of modifying it) might not be construed as distributing / publishing it.
The GPL was not designed to restrict interactions between an organisation and contractors doing bespoke coding for it. (And I doubt that FSF would attempt to enforce the GPL for that purpose.) The GPL is more concerned with leveling the playing field in interactions between software providers and their customers.
Anyway, I think this is academic. Every Government has secrecy laws that would make it
illegal for someone without the appropriate
clearance to view sensitive software. These laws would override contract law in any legal jurisdiction.
The GPL would allow the Government to release source-code for any changes to GPL'ed software made under constract. But it wouldn't require the Government to do this... unless the Government then supplies the modified software to a third party.
The situation is analogous to me contracting a developer to make some custom changes to a GPL'ed tool for my own personal use. If I stipulate (as a term of the contract) that I own it IP of the work done under the contract, then I am free to keep that IP (i.e. the modified source code) to
myself... provided I don't provide the binaries to someone else.
The GPL would allowrequire the Government to do this... unless the Government then supplies the modified software to a third party.
The situation is analogous to me contracting a developer to make some custom changes to a GPL'ed tool for my own personal use. If I stipulate (as a term of the contract) that I own it IP of the work done under the contract, then I am free to keep that IP (i.e. the modified source code) to
myself... provided I don't provide the binaries to someone else.
If there does not exist a Turing Machine (read: 'program') that can decide in finite time whether a given program meets a correctness criterion, then we can never decide whether a program is correct, because Turing Machines are the best computational tools we have. Therefore, suppose some Turing Machine M exists that accepts as input some suitable correctness criterion, and a program, and outputs whether that program meets the correctness criterion (in finite time).
I believe that you are misunderstanding Rice's theorem (etcetera) and its consequences in an subtle, but very important way. I'll try to explain this in terms of the Halting Problem, because I think that people are more familiar with this.
The Halting Problem is to find a program P that takes two inputs Px and Ix. Px is an arbitrary program for a Turing-complete language, and Ix is an arbitrary input set for Px. Our program P has to return "true" if Px will terminate for input Ix, and return "false" if it won't. And it has to return the result in a finite time.
The Halting Theorem says that the program P does
not exist. But what is more interesting here is what the theorem does NOT say. In particular, it does NOT preclude programs that will behave as
required for a significant subset of Px X Ix. It also does not preclude programs that will return
"true", "false" or "undecided" in a finite time. Finally, it says nothing about what proportion of the (infinite) space of Px/Ix combinations can (in theory) be proved to halt. Thus, we have the possibility that someone could write a very useful (though not universal) "halting tester" program.
Let's return to the bigger question of Rice's theorem and problem of proving programs correct.
The key is the meaning of "algorithmically decideable". AFAIK, this means that there are no "universal" algorithms that can prove a non-trivial property about ANY program in a finite time. But (as with the Halting Theorem), it does not say that useful algorithms cannot exist, or what proportion of the space of programs these algorithms can (in theory) operate over.
I hope that you realise that your notion that
"there's no absolute truth outside of abstract
thought" is itself a belief (or a supposition)
rather than an absolute truth.
And this is just symptomatic of the legal system's
broken-ness. The system allows so many legal tactics that have nothing to do with the real merits of a case, and nothing to do with the ideals of justice.
To make any object persist, you simply make it reachable.
What if a non-OOP language/tool needs access, especially query access, to the data?
There are lots of answers to that:
If your OPPL is an extension of a general
purpose programming language (e.g. Java), why
do you need to use another language? (OK, there
are cases, but this is the exception, not the rule. In the same way that it is exception to have to mix C / C++ with Java.)
The OPPL could support calls to/from embedded libraries written in another language, for example like JNI and the Invocation API do for Java.
The OPPL could provide a set of libraries that implement the notional equivalent of an SQL engine. Programs written in other languages could access the OPPL native data via the equivalent of JDBC / ODBC / whatever.
You could make your OPPL application a network server.
Which language? Application language? A language like Oracle's PL/SQL? If in Java, why is Java better than a relational language (like SQL, although SQL is not the best IMO)?
Would you write a compiler or a scientific application in PL/SQL? Why not?
Think of (say) a dialect of Java in which the way you accessed / updated data was completely independent of its persistency.
I am not sure what you mean by this.
In a typical OPPL (e.g. PSAlgol, Napier88, pJama), the only thing that distingusishes persistent and non-persistent data is that persistent data can be reached by following pointers / references from a distinguished "persistent root" object. To make any object persist, you simply make it reachable. To access a persistent object, you simply follow the pointers. The runtime system takes care of all of the rest, behind your back.
Databases are for more than just persistence. If you are only using it for persistence, then you are wasting it.
I've already pointed out that most of what a conventional DBMS does is also doable by the OPPL itself, or by libraries built therein. However, I take issue with your assertion that I'm "wasting it" if I only use a DBMS for simple persistence. This is often the best technical alternative available, and it is often what the customer demands. One reason for keeping your applications's use of the DBMS simple is to reduce problems porting another DBMS.
It differs in that it has all the functionality
/ usability of a conventional programming language,
including all of the data types / constructors and all of the control constructs you'd expect to see
there. Think of (say) a dialect of Java in which the way you accessed / updated data was completely independent of its persistency.
BTW, I didn't mention Java by accident. There are two existing implementations of Java with support for OP; i.e. 'pJama' and 'Kissme'. In both cases, no changes were made to the Java language to support OP. Rather, it is implemented (respectively) in the bytecode generator and in the Java virtual machine.
Relational tables impose certain constraints on data that OO does not. This might seem limiting at first, but it also allows the power of "relational algebra", which OO does not have a usable equivalent of yet.
I'd say that languages like OCL and OQL are OO's moral equivalent to a language for relational algebra. But there is still an impedence mismatch between OO query languages (e.g. OCL/OQL) on the one hand and OO programming languages (e.g. Java) on the other. What is needed is a language / environment that supports both paradigms within a single syntactic framework and type system/value domain.
Your description makes it sound like databases are *only* for persistence.
That was not my intention. Classical DBMS do a lot more besides simply making data persist. But most of the things you list as things that DBMS do could also be done in a fully fledged Orthogonally Persistent Programming Language / environment. For example:
Assuming that a query language and query optimizer is required, these could be implemented as a suite of library code that makes use of reflection.
Constraints and data validation could be handled the same way, or they could be handled by statically generated code modules, emitted by the programmer's IDE.
Contention management (transactions) should be part of the OPPL's core infrastructure; e.g. core libraries.
Ditto for security/access control, backup, replication and data migration.
I'm not saying that OPPL's currently support these things. But with enough work they could!
Slashdot Mirror
That explains why my fingers keep getting covered in saliva!! :-)
Personal bankruptcy is not an easy out ... so I've heard.
I'll pardon your naivety, but not your spelling :-)
(OK ... I don't think this is surprising at all. Perhaps you shouldn't either.)
I heard a radio interview with the baloon's developer on the BBC World Service last night. When the ballon got loose, it's battery got ripped off and hence its "brain" is "dead". (The developer made this very clear ... in response to a series of decidedly lame questions by the interviewer.) Any discussion of the "intelligence" (or otherwise) of this this particular balloon is moot.
Unfortunately, some spammers are using so called "reverse proxies" installed on hacked machines to host the webpages / email boxes mentioned in the crap they send you. To find the true address for the spammer you need to locate and reverse engineer the hacked machine.
Actually, this is common sense. If you wanted to damp combustion by excluding oxygen, it would be cheaper (and more environmentally friendly) to use an inert gas like nitrogen or carbon dioxide.
There is no doubt that Halon does replace oxygen to some degree and therefore does present a potential danger of asphyxia. However, there is another problem with Halons. When they come into contact with a fire, they breaks down, releasing breakdown products that are extremely dangerous, even at low concentrations.
No lawyers
If I remember correctly (from my 4.2bsd days), you can also run into trouble if your hard drives have bad blocks, and you revector them in the OS's file system driver. Doing a 'dd' on a raw partition would run into the bad blocks, and bad things may happen to your backup.
In this case, a good answer would be "Stop wasting time emailing your web page to your friends and get back to cramming for your exams!" :-)
However, if you offered to do a live demonstration at the US Patent Office HQ, I'm sure that you could get a patent granted :-)
Alternatively, (IANAL) the contractors might legally be construed as agents of the Government for the purposes of IP control wrt the contract. Hence, providing them with a copy of the modified software (for the purposes of modifying it) might not be construed as distributing / publishing it.
The GPL was not designed to restrict interactions between an organisation and contractors doing bespoke coding for it. (And I doubt that FSF would attempt to enforce the GPL for that purpose.) The GPL is more concerned with leveling the playing field in interactions between software providers and their customers.
Anyway, I think this is academic. Every Government has secrecy laws that would make it illegal for someone without the appropriate clearance to view sensitive software. These laws would override contract law in any legal jurisdiction.
To see what I meant to write, please read my other (corrected) reply to this post's grandparent. Sorry.
AFAIK, this is incorrect. (IANAL, etc)
The GPL would allow the Government to release source-code for any changes to GPL'ed software made under constract. But it wouldn't require the Government to do this ... unless the Government then supplies the modified software to a third party.
The situation is analogous to me contracting a developer to make some custom changes to a GPL'ed tool for my own personal use. If I stipulate (as a term of the contract) that I own it IP of the work done under the contract, then I am free to keep that IP (i.e. the modified source code) to myself ... provided I don't provide the binaries to someone else.
The GPL would allowrequire the Government to do this ... unless the Government then supplies the modified software to a third party.
The situation is analogous to me contracting a developer to make some custom changes to a GPL'ed tool for my own personal use. If I stipulate (as a term of the contract) that I own it IP of the work done under the contract, then I am free to keep that IP (i.e. the modified source code) to myself ... provided I don't provide the binaries to someone else.
I believe that you are misunderstanding Rice's theorem (etcetera) and its consequences in an subtle, but very important way. I'll try to explain this in terms of the Halting Problem, because I think that people are more familiar with this.
The Halting Problem is to find a program P that takes two inputs Px and Ix. Px is an arbitrary program for a Turing-complete language, and Ix is an arbitrary input set for Px. Our program P has to return "true" if Px will terminate for input Ix, and return "false" if it won't. And it has to return the result in a finite time.
The Halting Theorem says that the program P does not exist. But what is more interesting here is what the theorem does NOT say. In particular, it does NOT preclude programs that will behave as required for a significant subset of Px X Ix. It also does not preclude programs that will return "true", "false" or "undecided" in a finite time. Finally, it says nothing about what proportion of the (infinite) space of Px/Ix combinations can (in theory) be proved to halt. Thus, we have the possibility that someone could write a very useful (though not universal) "halting tester" program.
Let's return to the bigger question of Rice's theorem and problem of proving programs correct. The key is the meaning of "algorithmically decideable". AFAIK, this means that there are no "universal" algorithms that can prove a non-trivial property about ANY program in a finite time. But (as with the Halting Theorem), it does not say that useful algorithms cannot exist, or what proportion of the space of programs these algorithms can (in theory) operate over.
I hope that you realise that your notion that "there's no absolute truth outside of abstract thought" is itself a belief (or a supposition) rather than an absolute truth.
And this is just symptomatic of the legal system's broken-ness. The system allows so many legal tactics that have nothing to do with the real merits of a case, and nothing to do with the ideals of justice.
There are lots of answers to that:
Would you write a compiler or a scientific application in PL/SQL? Why not?
I am not sure what you mean by this.
In a typical OPPL (e.g. PSAlgol, Napier88, pJama), the only thing that distingusishes persistent and non-persistent data is that persistent data can be reached by following pointers / references from a distinguished "persistent root" object. To make any object persist, you simply make it reachable. To access a persistent object, you simply follow the pointers. The runtime system takes care of all of the rest, behind your back.
Databases are for more than just persistence. If you are only using it for persistence, then you are wasting it.
I've already pointed out that most of what a conventional DBMS does is also doable by the OPPL itself, or by libraries built therein. However, I take issue with your assertion that I'm "wasting it" if I only use a DBMS for simple persistence. This is often the best technical alternative available, and it is often what the customer demands. One reason for keeping your applications's use of the DBMS simple is to reduce problems porting another DBMS.
You're not Steve Blackburn are you? No :-)
It differs in that it has all the functionality / usability of a conventional programming language, including all of the data types / constructors and all of the control constructs you'd expect to see there. Think of (say) a dialect of Java in which the way you accessed / updated data was completely independent of its persistency.
BTW, I didn't mention Java by accident. There are two existing implementations of Java with support for OP; i.e. 'pJama' and 'Kissme'. In both cases, no changes were made to the Java language to support OP. Rather, it is implemented (respectively) in the bytecode generator and in the Java virtual machine.
I'd say that languages like OCL and OQL are OO's moral equivalent to a language for relational algebra. But there is still an impedence mismatch between OO query languages (e.g. OCL/OQL) on the one hand and OO programming languages (e.g. Java) on the other. What is needed is a language / environment that supports both paradigms within a single syntactic framework and type system/value domain.
That was not my intention. Classical DBMS do a lot more besides simply making data persist. But most of the things you list as things that DBMS do could also be done in a fully fledged Orthogonally Persistent Programming Language / environment. For example:
I'm not saying that OPPL's currently support these things. But with enough work they could!