 OZO Premium join:2003-01-17
| reply to modemslayer Re: [Chat] C++, Thinking out loud ...
said by modemslayer :Ahh the old days of BoundsChecker - You had suppression libaries to supress warnings of all the memory leaks in Windows, and all the memory leaks of the MS libraries, and third party tools, and so on, before you even got to the memory leaks in your own code. All of that goes away with .NET. I wonder how IE7 is written? C++, C#, VB?
Sometimes if I let IE7 just to keep it sitting on a Yahoo web site (or on some others too) for a day or two without any interaction I may discover that IE now consumes 700 - 900 MiB. Kill and restart is the only way to deal with beast. Anybody care now to check memory leaks or there is a new approach in town - suggestion to user: "buy more memory, it's so cheap now".
Or, time to time I may want to check memory allocations of Windows XP (pre SP 3) and if I compare the amount of memory allocated before I load my first program with memory allocated in a couple of days later when I unload all my programs from it (actually returning back to initial state), I may see 100, 200, or sometimes even 300 MB of difference.
What is it? A language used in development, sloppy developers using it, saving money on proper tests, or a new way of thinking in development in general?
In my projects using C++ all bytes were always counted, BTW. -- Keep it simple, it'll become complex by itself... |
|
 modemslayer
join:2003-12-11 Spokane, WA
·Comcast
| I know IE6 was written in C++, so I'm guessing they just moved the codebase forward with IE7/IE8. It would probably be a massive undertaking to re-write such a large browser in .NET, but I wonder...
I think it's not so much that developers are sloppy, it's just that you can take 10 developers at random and find a wide range of discipline between them. And the most disciplined C++ coder on earth is still going to dangle pointers and leak memory. Couple that with the philosophy that most project managers have that resources for unglamorous activities like leak detection, stress testing, diagnostics are to some degree wasted effort. There's no glory in leak detection. Never has been and never will be.
Now with stable, robust frameworks like Java and .NET, it's not necessary to devote your development resources towards these "unglorified" things. Your not having to go against human nature, which is "SHIP IT". With C# you can have all the glory and all the reliability. |
|
  nil Java Geek join:2000-11-27
Host: Webmasters and Dev.. Forum Feature Requ..
1 edit | reply to Steve said by Steve :said by Gomez : Steve doing C#?.. The difference between C# and Java is that Java sucks .. and while both languages are quickly becoming irrelevant outside of business-type programming, Java looks like it has a brighter future than C#.
Groovy is one language worth looking into, Scala is another.
That's only if you want to stay current on technologies, of course. Otherwise, there's always cobol. -- Life is too short to be boring |
|
  Steve I'm a PC, so shut up Consultant join:2001-03-10 Yorba Linda, CA
| said by nil : Otherwise, there's always cobol. I prefer the object-oriented version: ADD ONE TO COBOL
 |
|
  Gomez Been drinking brew for breakfast? Premium,Mod join:2001-02-21 Atlanta, GA clubs:
·Speakeasy
Host: Pub Games It is the economy,.. Clearwire
| said by Steve :said by nil : Otherwise, there's always cobol. I prefer the object-oriented version: ADD ONE TO COBOL Wouldn't that be ADD 1 TO COBOL? You give more credit to the parser than it deserves. -- It's a fact : Chicks dig Mafia players. 'Wanna help buy a goat?' - »www.kiva.org
|
|
  drew Reformation Premium join:2002-07-10 Port Orchard, WA clubs:
·wavebroadband
| reply to nil Here's my question: if Java and C# are going away for everything but desktop business, what's there for desktop consumer?
I say that because computing in the cloud, so far, sucks. Google Apps is by far the most prevalent and well known IMO, yet even in Chrome, which should be the *best* performer for Google Apps, it's dog slow. What happens when I want to work on a letter to grandma when I'm not able to get online? -- Come play Mafia! | My Picture Blog |
|
  nil Java Geek join:2000-11-27
Host: Webmasters and Dev.. Forum Feature Requ..
| There are many new languages coming up. A few are mentioned in this thread.
Let me put it this way:
10 years ago, enthusiastic geeks were using Java and business managers were scoffing, these days business managers want Java and enthusiastic geeks moved on.
If you want to stay relevant, you have to stay current. -- Life is too short to be boring |
|
  Gomez Been drinking brew for breakfast? Premium,Mod join:2001-02-21 Atlanta, GA clubs:
·Speakeasy
Host: Pub Games It is the economy,.. Clearwire
1 edit | reply to drew said by drew :Here's my question: if Java and C# are going away for everything but desktop business, what's there for desktop consumer? I say that because computing in the cloud, so far, sucks. Google Apps is by far the most prevalent and well known IMO, yet even in Chrome, which should be the *best* performer for Google Apps, it's dog slow. What happens when I want to work on a letter to grandma when I'm not able to get online? it's a matter of the best solution.. There is a paradigm shift happening, and those who are not noticing are in trouble.. Win32/.net/other current MS tech isn't going away.. It's firmly entrenched.
Web technologies are evolving rapidly. All the web technologies that were safe bets two years ago are out of favor now. the new hotness is ruby/rails, jruby/rails, groovy/grails, php/symphony..
Doesn't really matter if you are a couch hacker. But if it's a career, it's vital to adapt. -- It's a fact : Chicks dig Mafia players. 'Wanna help buy a goat?' - »www.kiva.org
|
|
  drew Reformation Premium join:2002-07-10 Port Orchard, WA clubs:
·wavebroadband
| reply to nil My apologies -- even after a quick re-read, I'm not seeing anything about new languages for the desktop. Only web techs.
@Gomez: I respect that the web is evolving VERY rapidly. We've had conversations about this and you've told me to get started on symfony, yesterday so to speak.
I'm personally concerned that there's too much movement towards the web. I can speak in detail about the problem due to experience in a product we're working right now. -- Come play Mafia! | My Picture Blog |
|
  Gomez Been drinking brew for breakfast? Premium,Mod join:2001-02-21 Atlanta, GA clubs:
·Speakeasy
Host: Pub Games It is the economy,.. Clearwire
| said by drew :My apologies -- even after a quick re-read, I'm not seeing anything about new languages for the desktop. Only web techs. @Gomez: I respect that the web is evolving VERY rapidly. We've had conversations about this and you've told me to get started on symfony, yesterday so to speak. I'm personally concerned that there's too much movement towards the web. I can speak in detail about the problem due to experience in a product we're working right now. (box + Brain).moveOutside
Any application that has human interaction can and should be web-based.. it's a common interface. Just as Chris Date said there is no reason that a relational database can't perform on par with a non-relational system many have have said the same about UIs. -- It's a fact : Chicks dig Mafia players. 'Wanna help buy a goat?' - »www.kiva.org
|
|
  drew Reformation Premium join:2002-07-10 Port Orchard, WA clubs:
·wavebroadband
| I *hate* web browsers though. Absolutely hate them.
None of them work as they should, there's no standardization.
Form design on a desktop app is not only significantly easier, it's far simpler to handle input from. Any time you design a web app, no matter if it's standards compliant or not, there's at LEAST one of the browsers you have to add in some kind of hack for.
I don't think my head's stuck in a box -- I'm open to new ideas. I'm just talking from a very low skill programmer's [insert word here for what I really am] perspective but mostly from a user's.
The only thing I've seen from a web technology that's really made me go "Whoa..." was Flex. Even then, it's not *that* cool. -- Come play Mafia! | My Picture Blog |
|
 modemslayer
join:2003-12-11 Spokane, WA
·Comcast
| reply to Gomez said by Gomez :Any application that has human interaction can and should be web-based.. For n-tier, client server based systems, sure...
But for pure desktop applications? I'm not buying that. You'd have desktop applications like photoshop and Nero be web-based? Acronis true image? AVG apps? Video games like World of Warcraft? What would be the benefit for a complex desktop application to run locally inside of a browser? IMHO you'd be setting desktop application development back, not moving it forward.
Adobe could make photoshop web-based. Blizzard could make WoW web-based. Why aren't they doing that? There's lots of web-based office productivity applications out there, yet people still like desktop apps like MS Office...
Maybe it's the same as arguing religion lol. |
|
  nil Java Geek join:2000-11-27
Host: Webmasters and Dev.. Forum Feature Requ..
| reply to drew That's where you will likely see most language development in coming years. Until there is some new technology to focus on.
My prediction: biggest areas of development in the next couple years will be cross-device computing. Think applications that can run on phones, TVs, game consoles, web and yes, desktops. -- Life is too short to be boring |
|
 dave Premium,MVM join:2000-05-04 not in ohio
·Verizon Online DSL
·Verizon FIOS
| reply to usa2k I had thought that the LINUX camp would still be evolving C++. Languages don't have to keep getting fatter and fatter and fatter. At some point, the language is good enough.
I basically won't use a non-standard feature, so if there's any advance in the language I use, it has to come through the standards committee.
Here's the current in-dev version »www.artima.com/cppsource/cpp0x.html (I assume C++ 0x isn't actually here yet - anyone?) |
|
 dave Premium,MVM join:2000-05-04 not in ohio
·Verizon Online DSL
·Verizon FIOS
| reply to modemslayer said by modemslayer :And the most disciplined C++ coder on earth is still going to dangle pointers and leak memory. I don't.
Don't blame the language if people have poor programming skills. |
|
  Steve I'm a PC, so shut up Consultant join:2001-03-10 Yorba Linda, CA
| said by dave :I don't. That's two of us.
There really is a disciplined way to avoid leaking memory and writing to memory you don't own. |
|
  Bobcat Premium join:2001-02-04 Bedminster, NJ
·Verizon Online DSL
| said by Steve :There really is a disciplined way to avoid leaking memory and writing to memory you don't own. In our systems, we allocate all the needed memory at initialization. We never deallocate memory. We never allocate memory after initialization. |
|
  Steve I'm a PC, so shut up Consultant join:2001-03-10 Yorba Linda, CA
| said by Bobcat :In our systems, we allocate all the needed memory at initialization. We never deallocate memory. We never allocate memory after initialization. That's a fine approach if the application domain permits it; many programs do not. |
|
 OZO Premium join:2003-01-17
1 edit | reply to dave said by dave :said by modemslayer :And the most disciplined C++ coder on earth is still going to dangle pointers and leak memory. I don't. Don't blame the language if people have poor programming skills. Or they're not applying a good, well established practices how to manage it, or they are not required to completely and permanently test their products, or, simply, they don't care (and why software development should be a job for anyone?) As I've said earlier, in my C/C++ projects all bytes are counted. So, don't blame programming language, it's who applies it makes the matter. -- Keep it simple, it'll become complex by itself... |
|
  Bobcat Premium join:2001-02-04 Bedminster, NJ
·Verizon Online DSL
| reply to Steve said by Steve :said by Bobcat :In our systems, we allocate all the needed memory at initialization. We never deallocate memory. We never allocate memory after initialization. That's a fine approach if the application domain permits it; many programs do not. I know. I'm talking about a different animal.
Two reasons we do it that way: 1) The obvious performance benefits of not allocating/deallocating memory on the fly. 2) We have to meet the worst-case* memory margins, so there's no real benefit of allocating/deallocating memory on the fly.
* Definition of worst-case design: The one delivered. 
-- It's nice to be important, but it's more important to be nice. |
|