that's fine if your program is so simple that every posible code path can be executed in a short amount of time, but for anything but the most trivial rograms this is impossible (see Turing for a reason why). This means that for dynamically typed languages, the onus is on testing to find a whole slew of errors that statically-typed languages can find at compile time.
not a good solution for multi-K-line programs written by large teams of programmers.
when the the right-hand-side expression is an anonymous type (or an array, collection of anonymously-typed objects) then there is non name you can write on the left.
if you're running windows servers, lock them down (both externally and internally), lock down your Active Directory.
If you want XP Home machines to be able to authenticate on the domain, just force them to connect to an internal VPN - their VPN credentials will be used for connections to local services (exchange, file servers, etc...)
wait, you're bitching about paying $100 for the DDK (and if you're writing drivers, then that should be about the same amount as you could earn in a week's toilet breaks) but at the same time you're saying that that cost hinders you from writing user-mode applications? I'm missing something there...
that's true, but there are ways around that, and the fact remains that many apps (Apache & MySql come to mind) are moving to an in-process concurrency model based on threads, primarily for the performance gains it provides.
pipes exist on windows just fine (named or otherwise). and while the unix fork call is elegant in its simplicity, it's mostly useless for real work since while the child process inherits almost everything from the parent, the first thing the child has to do is to uninherit all the stuff it doesn't want (like any sensitive data it doesn't want distributed around the process tree). if all you're doing is creating a bunch of identical processes then the modern (ooh!) usage of threads is usually a much better model.
just because it wasn't invented by a couple of guys in the '70s with a few K address space, doesn't mean it automatically sucks.
erm, WaitForMultipleObjects is essentially the same as select(), execpt it supports waiting on objects, not just sockets. sockets are waitable objects in NT, as are files, processes, threads, named-pipes, events, etc... garbage indeed.
of course, you should be using IOCompletionPorts instead...
not impossible, just hard (and not done yet). VS provides suggestions based on context: ie, which identifiers/types are currently in scope and/or relevant. the intellisense engine uses the same definition of scope that the compiler uses, and it does it across multiple projects, referenced libraries/typelibs, taking into account project build settings and a deep understanding of the semantics of the language. it's not your father's ctags
yup, and that's the same reason they're rated as a higher threat. there could be one single vulnerability for windows and another for linux and the windows one would be a more serious threat since so many more people would be affected.
Slashdot Mirror
not a good solution for multi-K-line programs written by large teams of programmers.
Your point is well taken, though. I think...
If you want XP Home machines to be able to authenticate on the domain, just force them to connect to an internal VPN - their VPN credentials will be used for connections to local services (exchange, file servers, etc...)
wait, you're bitching about paying $100 for the DDK (and if you're writing drivers, then that should be about the same amount as you could earn in a week's toilet breaks) but at the same time you're saying that that cost hinders you from writing user-mode applications? I'm missing something there...
dude, buy almost anything from newegg.com (or similar) and you can get an OEM XP for $50.
actually, the correct solution is to design to the standards, and as for complaining customers: shoot them in the name of open standards.
viva la revolution!!
<joke/>
that has to be the lamest excuse for a list of reasons why not to use something.
that's true, but there are ways around that, and the fact remains that many apps (Apache & MySql come to mind) are moving to an in-process concurrency model based on threads, primarily for the performance gains it provides.
they aren't guaranteed to be coherent in multi-processor situations. the same is true for Linux.
just because it wasn't invented by a couple of guys in the '70s with a few K address space, doesn't mean it automatically sucks.
yeah, or you could just call the CreatePipe or _pipe functions.
of course, you should be using IOCompletionPorts instead...
Yes, he's an idealist. But as soon as he opens his mouth he becomes something else as well...
Windows XP & Server 2003 load .NET .EXEs natively. the CLR loader stub is bypassed.
The big question of course is: does the WoD help society more that it hurts it?
if I steal 1,000,000 $1 bills do i get a longer sentance than if I stole 10,000 $100 bills?
not impossible, just hard (and not done yet). VS provides suggestions based on context: ie, which identifiers/types are currently in scope and/or relevant. the intellisense engine uses the same definition of scope that the compiler uses, and it does it across multiple projects, referenced libraries/typelibs, taking into account project build settings and a deep understanding of the semantics of the language. it's not your father's ctags
yup, and that's the same reason they're rated as a higher threat. there could be one single vulnerability for windows and another for linux and the windows one would be a more serious threat since so many more people would be affected.