hey, you set me up with a hardware grant from suse, redhat, et al, and I'll test every possible combination of hardware given me.
Granted, the site will have to support itself with porn or some other equally immoral pasttime to pay for all the electricity I won't be earning money for at my dad job, but that's okay, I can live with a porn studio in my house...
Even your option is fraught with potential error. A simple:
int b(Thing *ptr) {
auto_ptr<Thing> aptr(ptr);... }...
Thrown in carelessly screws the pooch. I would personally prefer a refcounted or encapsulated smart_ptr vs. an auto_ptr. I like to keep my pointers consistent. A rule such that you have requires knowing that ptr.get() is a managed pointer, so that if you have to resize or prematurely delete it, you're updating the manager properly. It's inconsistent, and fraught with peril if your working with numerous developers. Double deleting a pointer has always been an undefined C++ operation, no?
I'm not trying to bash your practice although it kind of looks like it. I'm only bashing auto_ptr.:-) You *ARE* right, if you always pass a smart_ptr by reference you eliminate the risk. It only takes one moron on your project to leave off that & in the function prototype to cause you a weeks worth of HELL, though.
Personally, I think auto_ptr only got added to the standard so that people could feel safe knowing that dynamically allocated memory would get freed during stack traversal on an exception. I don't think it was *EVER* intended to be used except as you described in your post.
Sorry, not bashing the parent, just bashing the mentality of the thought that Microsoft thought they could get anything out of this maneuver (if indeed they staged it), especially without evidence to back it up. It's hilarious when you right down to it. And even if there *WAS* evidence, other than specific legal action against the code contributors themselves, coupled with the fact that copyrighted code would be extracted at first opportunity, it's senseless what they thought they could buy by settling themselves in with SCO.
Unless of course, they want to set themselves up as one of SCO's biggest creditors, so in the final bankruptcy/dissolution proceedings, they get the sum of all Unix IP that SCO lays claim to, or it gets shared betwixt Microsoft and SUN.
I can EASILY see this being the sole prime reason they've gotten the large $8m+ investments considering they came about after the evidence was shown to be not forthcoming (if I remember my timetables correctly).
MMM.... not 13, but perhaps not as articulate as I could have been Friday.:-D
1. Your existing JVM doesn't break said 5 year old Java program. 2. Failing that, you can still find a compatible version of the 5 year old JVM.
Java is no panacea, my friend. I've had numerous production tools break when we upgraded to another version of the JVM. And with Sun NOT supporting 1.0.7 anymore, I'm not too assured about Sun's long term prospects from keeping my code working.
What's your point, that Java is a great language, or that bytecode VM's are a good thing? There's arguments for both, but I argue for the VM long before I'd argue that Java is a "Good Thing(TM)"
Please, name me a compiler that deals with big/little endian issues at compilation. I'm genuinely curious. I know a LOT of CORBA guys who'd be interested in an optimization like this.:-)
Why is that so bad? With CPU power getting so crazy these days, and the needs of most users far outstripping their usage, why shouldn't we go back to timesharing, in limited usage scenarios? This is a *GREAT* solution for my dad's small office where he needs to bill customers, and my mom needs to invoice the vendors, enter PO's.
Yeah, and hows that working out for you BillG? Hmmm??? I know more and more people who are deploying Linux, and one massive corporation looking at a company-wide rollout and could care less about your piddly SCO lawsuit...
Oh Geez, you sound like one of those Team-OS/2 wankers who got off on trying to prove OS/2 had more than a paltry user base.
Linux is successful, don't misunderstand me, but don't try and make up numbers for it. Like every good revolution, the incumbent powers underestimate the widespread dissention happening right in front of their very eyes.
Hunt? HUNT? What's so hard about "K button" -> Internet -> WWW -> Konqueror Browser ??
You one of those people who had trouble understanding "Start -> Programs" like my dad did?
Start button? eh sonny, what's that?
</humor>:-)
I'm not trying to convince you, convert you, or call you a troll (I'm not the GP poster). I'm just trying to debunk a flawed argument. The rest of your post stands on it's merits.;-)
But you can't. Some might go in this directory, some get spread all around your Documents and Settings folder, some *STILL* go into the windows folder, and you get registry pollution, such that if you want to share your computer with someone, your settings may conflict with their settings, and ruin your day.
I won't even go into.DLL hell. Thankfully with "file protection" and "system recovery" that seems to be a thing of the past.
Most of these people bashing Linux about this realize that the only reason it's done that way on Windows is because it's EXPENSIVE to get decent web editing tools, image manipulators and viewers on Windows. You *HAVE* to build everything in, and those people usually do a pisspoor job at integration. Oh, I can crop, but I can't mirror? ack!!!
There's no *NEED* for on does-it-all-and-the-kitchen-sink type application for Linux/X. It's just not necessary. The OSS world has given us fully functional versions of *ALL* of the above applications, that you can download and use for free (and even give to your friends!) and that get better every day.
Why should I use Zoombrowser or some other piece of crap when I have the gimp, nautilus and firefox, or whatever it's HTML editor is?
The whole mentality is different. And if getting all our applications back to integrated solutions is where we are heading, and losing the great flexibility we have today, then I say SCREW the desktop population, let them suffer or thrive on Windows as they choose, and let us alone.
Stop making Linux into Windows, and just keep making it the best computing platform available.
You could say the same about Microsoft and DOS 1.0.
And if IIRC, the LinuxOne IPO was one of the top 5 IPO's in history, and that's at the height of the insanity. Apple, Pfizer, IBM, Ford... none of them had a chance compared to that hype-fest.
> Even the distros with the slickest installers don't match what OSX or XP offer in this regard.
I call bullshit.:-)
--- I should back that up:
last night, for example, trying to install ogle. Firefox brought up dialogs asking me if I wanted to manage the RPMs with YAST to complete an install, verified all the dependencies, and installed it.
Slashdot Mirror
As if vendors provide support these days without a $1000 yearly contract or $100/incident paid support options.
It is if it causes your machine to crash.
^dad^day^
hey, you set me up with a hardware grant from suse, redhat, et al, and I'll test every possible combination of hardware given me.
Granted, the site will have to support itself with porn or some other equally immoral pasttime to pay for all the electricity I won't be earning money for at my dad job, but that's okay, I can live with a porn studio in my house...
No, I think you're confusing a hummer IN a Hummer. :-)
Even your option is fraught with potential error. A simple:
... ...
:-) You *ARE* right, if you always pass a smart_ptr by reference you eliminate the risk. It only takes one moron on your project to leave off that & in the function prototype to cause you a weeks worth of HELL, though.
int b(Thing *ptr) {
auto_ptr<Thing> aptr(ptr);
}
Thrown in carelessly screws the pooch. I would personally prefer a refcounted or encapsulated smart_ptr vs. an auto_ptr. I like to keep my pointers consistent. A rule such that you have requires knowing that ptr.get() is a managed pointer, so that if you have to resize or prematurely delete it, you're updating the manager properly. It's inconsistent, and fraught with peril if your working with numerous developers.
Double deleting a pointer has always been an undefined C++ operation, no?
I'm not trying to bash your practice although it kind of looks like it. I'm only bashing auto_ptr.
Personally, I think auto_ptr only got added to the standard so that people could feel safe knowing that dynamically allocated memory would get freed during stack traversal on an exception. I don't think it was *EVER* intended to be used except as you described in your post.
Cheers.
Sorry, not bashing the parent, just bashing the mentality of the thought that Microsoft thought they could get anything out of this maneuver (if indeed they staged it), especially without evidence to back it up. It's hilarious when you right down to it. And even if there *WAS* evidence, other than specific legal action against the code contributors themselves, coupled with the fact that copyrighted code would be extracted at first opportunity, it's senseless what they thought they could buy by settling themselves in with SCO.
:-D
Unless of course, they want to set themselves up as one of SCO's biggest creditors, so in the final bankruptcy/dissolution proceedings, they get the sum of all Unix IP that SCO lays claim to, or it gets shared betwixt Microsoft and SUN.
I can EASILY see this being the sole prime reason they've gotten the large $8m+ investments considering they came about after the evidence was shown to be not forthcoming (if I remember my timetables correctly).
MMM.... not 13, but perhaps not as articulate as I could have been Friday.
So what's changed in the past 5 years since I've used C++ extensively that I should be paying attention to?
#define WIN32_LEAN_AND_MEAN
#include <windows.h>
You'll dump a lot of the GDI and OLE stuff.
Yes, that's if:
1. Your existing JVM doesn't break said 5 year old Java program.
2. Failing that, you can still find a compatible version of the 5 year old JVM.
Java is no panacea, my friend. I've had numerous production tools break when we upgraded to another version of the JVM. And with Sun NOT supporting 1.0.7 anymore, I'm not too assured about Sun's long term prospects from keeping my code working.
What's your point, that Java is a great language, or that bytecode VM's are a good thing? There's arguments for both, but I argue for the VM long before I'd argue that Java is a "Good Thing(TM)"
Cheers.
Please, name me a compiler that deals with big/little endian issues at compilation. I'm genuinely curious. I know a LOT of CORBA guys who'd be interested in an optimization like this. :-)
One character is all that separates auto_ptr from being a horrible nuisance:
... ...
;-)
class Thing { };
int a(auto_ptr<Thing> abc) {
}
int b(auto_ptr<Thing>& abc) {
}
int main(...) {
auto_ptr<Thing> ptr(new Thing);
a(ptr);
b(ptr);
}
A simple typo, but one that *WILL* result in a NULL pointer exception.
Can you guess what the typo is?
The problem is that most people are used to running quicken, microsoft works, internet explorer, and solitaire (among a myriad of other games).
Works and Explorer OSS has an answer for, Quicken is the iffy one.
Please remember that free($$$) != free(liberty), and while free($$$) is great, free(liberty) is infinitely better.
Why is that so bad? With CPU power getting so crazy these days, and the needs of most users far outstripping their usage, why shouldn't we go back to timesharing, in limited usage scenarios? This is a *GREAT* solution for my dad's small office where he needs to bill customers, and my mom needs to invoice the vendors, enter PO's.
Yeah, and hows that working out for you BillG? Hmmm??? I know more and more people who are deploying Linux, and one massive corporation looking at a company-wide rollout and could care less about your piddly SCO lawsuit...
YAWN.
Oh Geez, you sound like one of those Team-OS/2 wankers who got off on trying to prove OS/2 had more than a paltry user base.
Linux is successful, don't misunderstand me, but don't try and make up numbers for it. Like every good revolution, the incumbent powers underestimate the widespread dissention happening right in front of their very eyes.
Cheers.
I love that, Linux, LotDT (Lord of the Desktop).... hehe. Small minds, boring Friday afternoon...
Hunt? HUNT? What's so hard about "K button" -> Internet -> WWW -> Konqueror Browser ??
:-)
;-)
You one of those people who had trouble understanding "Start -> Programs" like my dad did?
Start button? eh sonny, what's that?
</humor>
I'm not trying to convince you, convert you, or call you a troll (I'm not the GP poster). I'm just trying to debunk a flawed argument. The rest of your post stands on it's merits.
But you can't. Some might go in this directory, some get spread all around your Documents and Settings folder, some *STILL* go into the windows folder, and you get registry pollution, such that if you want to share your computer with someone, your settings may conflict with their settings, and ruin your day.
.DLL hell. Thankfully with "file protection" and "system recovery" that seems to be a thing of the past.
I won't even go into
Cuz sometimes, someone elses used shit is nicer than the new shit you could afford. :-)
Plus I hear that in many cities it's a serious PITA to move big, nice furniture, so it's more commonplace to sell/lease the place furnished.
Most of these people bashing Linux about this realize that the only reason it's done that way on Windows is because it's EXPENSIVE to get decent web editing tools, image manipulators and viewers on Windows. You *HAVE* to build everything in, and those people usually do a pisspoor job at integration. Oh, I can crop, but I can't mirror? ack!!!
There's no *NEED* for on does-it-all-and-the-kitchen-sink type application for Linux/X. It's just not necessary. The OSS world has given us fully functional versions of *ALL* of the above applications, that you can download and use for free (and even give to your friends!) and that get better every day.
Why should I use Zoombrowser or some other piece of crap when I have the gimp, nautilus and firefox, or whatever it's HTML editor is?
The whole mentality is different. And if getting all our applications back to integrated solutions is where we are heading, and losing the great flexibility we have today, then I say SCREW the desktop population, let them suffer or thrive on Windows as they choose, and let us alone.
Stop making Linux into Windows, and just keep making it the best computing platform available.
Thank you
</soapbox>
You could say the same about Microsoft and DOS 1.0.
And if IIRC, the LinuxOne IPO was one of the top 5 IPO's in history, and that's at the height of the insanity. Apple, Pfizer, IBM, Ford... none of them had a chance compared to that hype-fest.
> Even the distros with the slickest installers don't match what OSX or XP offer in this regard.
:-)
I call bullshit.
---
I should back that up:
last night, for example, trying to install ogle. Firefox brought up dialogs asking me if I wanted to manage the RPMs with YAST to complete an install, verified all the dependencies, and installed it.
> Even the distros with the slickest installers don't match what OSX or XP offer in this regard.
:-)
I call bullshit.