That's what I was thinking, sort of. The French guy probably knew, but either had his own motives, or he was told but is now kept out of the limelight because retrospectively, that would have been illegal. Don't forget that the western secret services do work together at certain levels, in spite of their respective governments' apparent disagreements (a lot of valuable intel during the second Iraq war came from the French and German intelligence agencies, in spite of their governments' official opposition to it, for example). Sometimes western governments play good-cop-bad-cop.
I think it's fair to say (and I would like to put it forward as a theory) that human beings, given almost no knowledge on the subject of (for example) astronomy (as happened time and again before, say, the 15th century - warning: European POV here), will yet almost certainly, during a lifetime, invent all sorts of things related to it; sundials, longest-day-discovery-things (and therefore yearlength-discovery-things). Perhaps it's possible to define a certain 'unschooled but using maximum concentration' benchmark of human intelligence based on these things.
I think if you gave fermilab $20KK and told them to come up with this new lamp though, that you would be certain of the outcome. Whereas now, it's up in the air a bit.
Oh, you can do most of it, except for having a genuine object model. You know, one that is recognizable by your fellow programmers. And I don't mind that much at all - I'm an avid C hacker myself. I just find that a lot of people around me, say 'C' when they mean 'C++', and they would be lost without an object model. One more question: inheritance ? What do you consider inheritance in C ? Because I think it to mean gratuitous casting to one of many superclasses, for example, and I can't really see how one would go about that in C.
Not only BerkeleyDB, but there's also GDBM, NDBM et cetera. People forget these things and start reinventing the wheel, apparently. Although I/do/ think that now Oracle has acquired BerkeleyDB, there should come a new, completely independent transaction driven DBM library in C. But that's just me.
The Burberry brand has been ruined in the UK by 'chavs' (that's what they call white trailer trash (minus the trailers) in the UK) all of a sudden discovering it. Even though the price of it hasn't changed (expensive stuff), the people will just buy it, no matter what. The moral of the story: no matter how hard you try to be a luxury brand, you have to always be prepared to be catapulted to where you don't want to be because of the market's whims. In that light, it helps if you have more sticks in the fire.
All these solutions being floated here don't take into account practicality; if I am a European business person who flies into JFK every other week, carrying a laptop full of company secrets - hell yes I would have the contents of my harddrive encrypted. And I would NOT reveal the password to anyone; the NSA is only too prone to pass interesting stuff on to US companies. Up- and downloading stuff onto servers elsewhere is simply no solution (for both security and practicality), and neither is giving up the password. What are my alternatives - how can I do business in the US now ?
>> the evidence that Jesus walked over water is exactly as strong as the evidence Julius Caesar conquered Gaul.
Eh, what are you smoking ? We have a three gospels that are mere copies of each other saying that jesus walked on water stated in a clearly aphoristic way, versus a verifiable autobiography ('de bello gallico'), a noteworthy (more or less) historian (Plutarch), contemporaries Sallust and the poet Catullus, and one of his worst political enemies (Cicero). What more do you want ?!
I know that. But this isn't just a case of Estonians doing something - I applaud them for wanting to defend every one of their borders. It's a case of NATO being an undiplomatic elephant here. I just wonder who is making these policy decisions, what goes on in their head and what their aims are. Last I checked, NATO was a military defence organisation, not an IT outsourcing company.
And monkeyboy shouting: Brewers Brewers Brewers ! And Bill Gates saying: 640 microliters ought to be enough for everybody. And an irritating alcohol-hole at always the same position in the can. And a EULA before you open the can (opening it, is agreeing with it) - no dispensing information about the taste or colour of it. And being only allowed to store it in an approved fridge. Ah what the hell.
It's all to do with the old perspective of having only small directories.
If you're using a stream, that means that someone will have to fill a pipe/piece of memory somewhere at the call to opendir(). That works, but starts to have an impact when you have very large directories. Also, you risk seeing 'ghost' entries; entries that have been deleted in between the call to opendir() and readdir(). You can defend that by saying it's a transaction; the user 'sees' the directory as it was at the time of creation of the cursor (your DIR*). But then again, that would be only useful if you could do other things during the same transaction. And there is no such mechanism in UNIX.
In the mean time, the whole issue could be fixed by saying that when there are changes to the structure of a directory during an iteration of it, you forfeit your cursor (as you would in any other database like system outside a transaction). Essentially, you would flag the DIR* to become useless. On top of that, of course, you would like to have an exclusive lock on a directory, so you can make sure that those changes can't be made. Either that, or you have to make your filesystem fully transactional.
Sorting is another thing; at opendir() the user must be able to choose offset, limit, field to sort on and search criteria. The WIN32 API went this way a bit, but it didn't go far enough. To expect userland to gobble, match, sort and paginate when the kernel can do it so much better - well, it works in a buffered way for small directories, as I said. But if we want true database-like capabilities on our filesystem (and I know I do) then the spec must be changed. Since all of the calls involved in this refer to process-global resources, you wouldn't necessarily have to change the C-API, you could just extend it.
Nah - the beef with java isn't that it's slow. Well, that too, but it's more in the interfacing; want to do something in C in between all your java - use the JNI if you dare ! Want to fire off a shell - see you thirty statements later ! Want to create a PDF ? Well, see, first we instantiate a worker-factory of this XML-class, and then we feed it through this parser-generator, and then we might end up with some PDF... on mondays. Want to stat a file, fork a process, set up a non-standard network communication, access shared memory ? Tough ! This java, man ! And it's in a world of its own. It's no miracle the java guys are always so evangelical - they're not *allowed* to see what's outside their world.
Slashdot Mirror
That's what I was thinking, sort of. The French guy probably knew, but either had his own motives, or he was told but is now kept out of the limelight because retrospectively, that would have been illegal. Don't forget that the western secret services do work together at certain levels, in spite of their respective governments' apparent disagreements (a lot of valuable intel during the second Iraq war came from the French and German intelligence agencies, in spite of their governments' official opposition to it, for example). Sometimes western governments play good-cop-bad-cop.
I think it's fair to say (and I would like to put it forward as a theory) that human beings, given almost no knowledge on the subject of (for example) astronomy (as happened time and again before, say, the 15th century - warning: European POV here), will yet almost certainly, during a lifetime, invent all sorts of things related to it; sundials, longest-day-discovery-things (and therefore yearlength-discovery-things). Perhaps it's possible to define a certain 'unschooled but using maximum concentration' benchmark of human intelligence based on these things.
Because, of course, ISPs could also forge legitimate looking TCP RST packets.
I think if you gave fermilab $20KK and told them to come up with this new lamp though, that you would be certain of the outcome. Whereas now, it's up in the air a bit.
Oh, you can do most of it, except for having a genuine object model. You know, one that is recognizable by your fellow programmers. And I don't mind that much at all - I'm an avid C hacker myself. I just find that a lot of people around me, say 'C' when they mean 'C++', and they would be lost without an object model. One more question: inheritance ? What do you consider inheritance in C ? Because I think it to mean gratuitous casting to one of many superclasses, for example, and I can't really see how one would go about that in C.
I don't mean to troll, but if you're mainly a C hacker, what good is your opinion on any object model ? You didn't accidentally mean C++, did you ?
NOBODY expects that knives fashioned out of soda cans !
What's wrong with my patent on 'the girlfriend' ?
In an environment where SOAP manages 900 calls per second, _anyone_ can score 50000 calls per second. No matter what your protocol is like.
Nobody wants to miss out on the VM market. We're slowly moving into mainframe territory with our chips. Even Apple will not want to miss it.
Yeah, just because a land rush on the south pole might be just what we need to get things going again. Damn libertarians.
- You can already install VMWare and install Mac OS X on that.
- Support for non-apple-kissed hardware running Mac OS X could come from another party.
VMs are killing any sort of legality of hardware/software binding license issues, fast. As both MS and Apple are discovering.
Not only BerkeleyDB, but there's also GDBM, NDBM et cetera. People forget these things and start reinventing the wheel, apparently. Although I /do/ think that now Oracle has acquired BerkeleyDB, there should come a new, completely independent transaction driven DBM library in C. But that's just me.
Well, is the stock all sold out ?
The Burberry brand has been ruined in the UK by 'chavs' (that's what they call white trailer trash (minus the trailers) in the UK) all of a sudden discovering it. Even though the price of it hasn't changed (expensive stuff), the people will just buy it, no matter what. The moral of the story: no matter how hard you try to be a luxury brand, you have to always be prepared to be catapulted to where you don't want to be because of the market's whims. In that light, it helps if you have more sticks in the fire.
All these solutions being floated here don't take into account practicality; if I am a European business person who flies into JFK every other week, carrying a laptop full of company secrets - hell yes I would have the contents of my harddrive encrypted. And I would NOT reveal the password to anyone; the NSA is only too prone to pass interesting stuff on to US companies. Up- and downloading stuff onto servers elsewhere is simply no solution (for both security and practicality), and neither is giving up the password. What are my alternatives - how can I do business in the US now ?
>> the evidence that Jesus walked over water is exactly as strong as the evidence Julius Caesar conquered Gaul.
Eh, what are you smoking ? We have a three gospels that are mere copies of each other saying that jesus walked on water stated in a clearly aphoristic way, versus a verifiable autobiography ('de bello gallico'), a noteworthy (more or less) historian (Plutarch), contemporaries Sallust and the poet Catullus, and one of his worst political enemies (Cicero). What more do you want ?!
Ok. White chocolate isn't chocolate, but fat and sugar. Now what do _you_ think ?
I know that. But this isn't just a case of Estonians doing something - I applaud them for wanting to defend every one of their borders. It's a case of NATO being an undiplomatic elephant here. I just wonder who is making these policy decisions, what goes on in their head and what their aims are. Last I checked, NATO was a military defence organisation, not an IT outsourcing company.
Doesn't sound at all like another attempt by NATO to encroach upon what Russia considers her former sphere of influence. Not at all.
It's not like I'm some sort of advocate of Russian politics, but someone inside NATO must have a clue about these things and does them deliberately.
Weird.
And monkeyboy shouting: Brewers Brewers Brewers ! And Bill Gates saying: 640 microliters ought to be enough for everybody. And an irritating alcohol-hole at always the same position in the can. And a EULA before you open the can (opening it, is agreeing with it) - no dispensing information about the taste or colour of it. And being only allowed to store it in an approved fridge. Ah what the hell.
Did the filesystem kill his wife ?
It's all to do with the old perspective of having only small directories.
If you're using a stream, that means that someone will have to fill a pipe/piece of memory somewhere at the call to opendir(). That works, but starts to have an impact when you have very large directories. Also, you risk seeing 'ghost' entries; entries that have been deleted in between the call to opendir() and readdir(). You can defend that by saying it's a transaction; the user 'sees' the directory as it was at the time of creation of the cursor (your DIR*). But then again, that would be only useful if you could do other things during the same transaction. And there is no such mechanism in UNIX.
In the mean time, the whole issue could be fixed by saying that when there are changes to the structure of a directory during an iteration of it, you forfeit your cursor (as you would in any other database like system outside a transaction). Essentially, you would flag the DIR* to become useless. On top of that, of course, you would like to have an exclusive lock on a directory, so you can make sure that those changes can't be made. Either that, or you have to make your filesystem fully transactional.
Sorting is another thing; at opendir() the user must be able to choose offset, limit, field to sort on and search criteria. The WIN32 API went this way a bit, but it didn't go far enough. To expect userland to gobble, match, sort and paginate when the kernel can do it so much better - well, it works in a buffered way for small directories, as I said. But if we want true database-like capabilities on our filesystem (and I know I do) then the spec must be changed. Since all of the calls involved in this refer to process-global resources, you wouldn't necessarily have to change the C-API, you could just extend it.
Free Kevin !
Nah - the beef with java isn't that it's slow. Well, that too, but it's more in the interfacing; want to do something in C in between all your java - use the JNI if you dare ! Want to fire off a shell - see you thirty statements later ! Want to create a PDF ? Well, see, first we instantiate a worker-factory of this XML-class, and then we feed it through this parser-generator, and then we might end up with some PDF... on mondays. Want to stat a file, fork a process, set up a non-standard network communication, access shared memory ? Tough ! This java, man ! And it's in a world of its own. It's no miracle the java guys are always so evangelical - they're not *allowed* to see what's outside their world.