I think of 1 application where C is better than Perl for file parsing type roles, strangely enough. That's for parsing web server log files and generating pretty stats pages.
Most people have used Webalizer, written in C, but move to AWstats, written in Perl. AWstats is prettier and has been updated more often - probably because its written in perl. However, Webalizer is a lot faster, so much so that the big hosting companies don't use the perl generators, they use ones written in C.
So, I think Perl has its place parsing files, but as usual, C wins if you spend the programmer cost (ie. take the time) to write your routines in it. I guess this should be best practise in the computer industry - write your performance critical bits in C, and the less-intensive parts in a scripting language.
maybe the application its put to doesn't equate to 'mission critical', but I find it has more scalability than many other systems, is more responsive under extreme load, and has almost zero downtime. I don't know the last time/. wasn't available when I visited the site.
So although its not a mission critical app, that doesn't mean the underlying technology isn't suitable. Far from it.
Why would you want to rewrite to use.NET, I mean c'mon Perl programmers are known for their objectivity and pragmatism. Rewrite in.NET before you *have* to, forget it!
There's 2 things to consider before you go changing your code:
1. COM may be 'oh, that old thing we no longer talk about' to Microsoft, but it isn't going away anytime soon, no matter what their marketing department tells you. There's a fair bit of code written that uses it.
2. One of those codebases that is heavily reliant on COM (and Win32) is this.NET thing, a lot of the class library is a wrapper around the old libraries. So even if you did rewrite your code, all you'd be doing is calling your old libraries through a intermediate layer!
Sure MS doesn't want to do IronPerl, I think that's because python and Ruby are 'cool' languages, and MS is trying to be like someone's Dad, 'getting hip with the kids'. I doubt it'll ever create an IronPerl simply because there's no mileage in it for them to entice the Perl developers over to Windows unlike the Python and Ruby folks that they're scared of losing to other platforms.
that's interesting.. I knew C# 3wasn't submitted to the standard, so LINQ and WPF aren't standards, but I didn't realise MS had added their own extensions to the standard already!
its not so much zealotry, if you code for Windows, you code using Microsoft libraries and everything that comes with Visual Studio. If it doesn't come with VS (or the platform SDK perhaps) then you don't use it.
I'm serious, things used to be a bit different, you could grab code from all over and use it without problem, but I've seen a definite shift away from that over the last few years.
Perhaps its because MS has supplied more library code, or perhaps its because of the Open source movement (ie managers have discovered people using free code, have read "free code is communist" on a website or magazine and have put 2 and 2 together to end up with "no use of free code in this organization").
So, nowadays, if you're going to write for Windows, you are very much pushed towards using the 'standard' or default libraries, and that means DirectX in this case.
OpenGL does have good libs, I like it, I've used it, but the number of Windows developers who'll use it is very small indeed. Its a shame and its why my attitude towards MS has shifted over the last few years, I am still a Windows developer, but I see the monoculture has grown too much.
More control means nothing, its all about the budgets.
If you have a budget, you buy as many windows licences as you need and you don't think twice about it (I'm being slightly simplistic of course). If your budget is cut, but you still must get the work done, you think "I wish I could get all those servers running without any licencing"... and you start to consider Linux.
There's a lot of people out there who "won't get fired for buying Microsoft", but not they cannot buy MS because they have no cash to spend, but they still have all that work still to do. What's a poor manager to do?
Then there's the technical route - you need to do something, you can't get a new VM installed (no Windows licence), so you install Linux and see... and perhaps you like it, and then you have 2 and 3 and pretty soon all your VMs apart from a couple of old legacy ones are running linux.
Maybe FOSS isn't cheaper overall, but it doesn't matter - its cheaper where it matters, in the upfront costs that the bean counters are counting. God knows there's enough inefficiency in IT (look at me, I'm writing design documents that pretty much simply rephrase everything in the requirement documents for no more than political reasons).
J2EE v.NET is not the same comparison. Java v Net is.
Does it really scale better? Or is it that no-one notices any more because computers have got more RAM and faster CPUs nowadays.
the API is (well, was) almost entirely the old Java/J++ class library.
Now that said, you're still right. MS *is* better at development software than Sun. Visual Studio is best in its class and.NET is easier to use mainly because of that. I wouldn't like to type out those long, long namespaced methods without the IDE to autocomplete for me!
well, it snot that decent a little language. Its a MS port of Java with extra bits added mainly in the realm of the GUI and interop with legacy COM and Win32.
They have added extra features that are cool, like LINQ, but I feel they'll be heavily misused over the coming years. They've also adding features like extension functions which will make scripting languages look like statically strong-typed languages compared to C# as time goes by.
Its also pretty complex for a little language, I'm seeing a lot of people complaining on the web about how they're using up all their memory followed by little tutorials on Dispose and releasing objects so they actually get disconnected and then freed by the GC.
I think if it wasn't for curly braces, and the.NET class library, hardly anyone would be using it. Imagine if MS only released VB.NET, you wouldn't use it, the classic VB crowd wouldn't use it, but its practically the same language as C#.
BTW. C#2 is a standard, C#3 is not. MS has promised to submit it, but they havn't yet.
really? I thought MS has dropped support for OpenGL, apart from its very legacy v1.1 base. And you know they will never write a managed wrapper for it.
So, who will write a.NET game using opengl? If you're a Windows dev (and most game devs are) then you'll be using DirectX.NET, which oh so conveniently is not available for Linux.
So close. Yet so immensely far. Do you see the problem now? Its business a usual for MS, but with the added bonus of saying "But we are working on Linux interoperability, Mr DoJ, look - Mono".
perhaps you need to give away free copies of a Linux magazine with every netbook sold?
People who buy Windows often don't know what they're doing with them anyway, give eg Linux mag away, and you'll get a lot less returns and happier customers.
on thew other hand, itr would be acceptable to sign a 'never compete ever ever again' contract, if they were prepared to pay you for the duration of that contract.
Most contracts that do this have limited non-compete clauss, like selling a shop, you promise not to open another one in the same town for 6 months to a year. After that though, you can do what you like. Such terms are written into the contract, Id say they were unenforceable otherwise.
Open source means the source is publically available.
and the Windows source code itself, if you sign all the legal NDAs in their 'shared source' program (and they let you, of course). So I guess you think Windows is an Open source program?
Nice. You've just destroyed the common-usage of the term. You'll be telling us 'gay' means happy and has no homosexual meaning next?
Words change their meaning over time and with majority usage. "Open source" nowadays refers to a particular system of free and shareable code. That's just the way its gone, you can try and change it, but the majority of people have to agree and use it in your terms for it to actually change like you want.
only to your mind, not to that of a business executive who feels the need to protect his company's intellectual property and maximise further corporate innovation through aggressively pre-empting market share downward factors to further enhance shareholder value. (and his bonus).
what you or I think should happen is nothing when it comes to making money.
Do you expect C#3.0 to become an ECMA and ISO standard, as previous versions have?
We're certainly open to that. There's no ongoing work in the standards committee at the moment, but it's really more a question of whether the community of industry partners out there would like to continue with that process. I should also say that the standards for C# explicitly do permit implementers to have extensions to the language, so though C# 3.0 is not standardized, it is certainly a complete implementation of the C# 2.0 standard. It is 100% backwards compatible, as all versions are.
Re:a bunch of questions
on
C# In-Depth
·
· Score: 1
no, if MS said No to Mono, those guys would bitch (quietly) about it and then start blogging about how dangerous it was to use an unsupported platform on an unstable operating system that did not have the quality integration of a well-tested stack of applications that only Microsoft can provide.
They'll do it to save you from yourself.
MS wouldn't care.
(from TFA: "It is possible to build alternate implementations. We are not building.NET for Linux, because the value proposition that we can deliver to our customers is a complete unified and thoroughly tested package, from the OS framework to databases to Web servers etc.")
Re:2001 called and wants its search engine back.
on
Google, Circa 2001
·
· Score: 1
actually, Microsoft called, said it'd like to go back to 2001 to try to take over the world some more:
On the failure to realise early enough the importance of internet search, Mr Ballmer said: "Do I wish we'd started the investment in search a few years earlier? Yes."
"We may be the David up against Goliath but we're working on it."
And he added that the real concern was the lead that Google had built up in online advertising.
"We probably missed the power of the advertising model, not so much the technology," Mr Ballmer said.
See, things have changed, once they'd have claimed it was the technology, honest. Now they don't give a fig about the tech - just the cash.
Re:remember the carnagie melon self guided vehicle
on
C# In-Depth
·
· Score: 1
Actually, they did fix the problem... since it was an issue with their code and not C#. C# was doing what it was supposed to.
and in related news, C++ is not hard, its just that you're not intelligent enough to use it. And javascript is not slow, its just that you're using it for applications with incorrect performance requirements, and java isn't memory-hungry, you're just running it in a constrained environment:-)
If you say that C# is working fine, its a problem with the coders, then my points above are equally true.
I've been looking through stackoverflow recently, and there are a lot of questions along the line of "my.net app uses too much memory", its one of the inherent problems of a GC architecture that you need to be aware of, just like C#'s other language features. I believe its a big issue that needs more visibility - as you say, its not the GC that's at fault, its the whole understanding of how to properly code in a GC environment.
Unfortunately, most.NET coders are drawn to the language because it has features like this that are designed to make it easier for them to code, so they don't usually want to know that it is harder than they expected.
Re:One of the most widely used languages?
on
C# In-Depth
·
· Score: 1
apart from the dodgy replication (new '911' ambulance customer, 2 sites, SQL 2005, clustered DB... support engineer called out at 2am because 1 DB had fallen over and the other one couldn't be bothered to take over)
Microsoft later shipped us a patch, which was nice of them, its just a pity that the latest rollup (v6) didn't contain the patches contained in patch v5.
Usually SQL Server problems are down to its locking strategy. Anyone coming from an Oracle background to work with SQL Server can easily get a totally unscalable DB. Our old product had to be extensively redesigned to work with SQLServer because of this 1 issue. 2008 apparently supports row-level locking much better so perhaps that's what he means by "getting it right" at last.
Re:a bunch of questions
on
C# In-Depth
·
· Score: 1
isn't that an indictment of their legacy development practices. If I look at all the projects I have here, I reckon I could consolidate a dozen into 1 very easily. I might, one day, get to merge the 3 almost-identical extract applications into just 1.
Now his app runs in 2 hours, if he'd refactored all that C++ code into a single C++ app, it'd probably run in 10 minutes:-)
Slashdot Mirror
I think of 1 application where C is better than Perl for file parsing type roles, strangely enough. That's for parsing web server log files and generating pretty stats pages.
Most people have used Webalizer, written in C, but move to AWstats, written in Perl. AWstats is prettier and has been updated more often - probably because its written in perl. However, Webalizer is a lot faster, so much so that the big hosting companies don't use the perl generators, they use ones written in C.
So, I think Perl has its place parsing files, but as usual, C wins if you spend the programmer cost (ie. take the time) to write your routines in it. I guess this should be best practise in the computer industry - write your performance critical bits in C, and the less-intensive parts in a scripting language.
maybe the application its put to doesn't equate to 'mission critical', but I find it has more scalability than many other systems, is more responsive under extreme load, and has almost zero downtime. I don't know the last time /. wasn't available when I visited the site.
So although its not a mission critical app, that doesn't mean the underlying technology isn't suitable. Far from it.
Why would you want to rewrite to use .NET, I mean c'mon Perl programmers are known for their objectivity and pragmatism. Rewrite in .NET before you *have* to, forget it!
There's 2 things to consider before you go changing your code:
1. COM may be 'oh, that old thing we no longer talk about' to Microsoft, but it isn't going away anytime soon, no matter what their marketing department tells you. There's a fair bit of code written that uses it.
2. One of those codebases that is heavily reliant on COM (and Win32) is this .NET thing, a lot of the class library is a wrapper around the old libraries. So even if you did rewrite your code, all you'd be doing is calling your old libraries through a intermediate layer!
Sure MS doesn't want to do IronPerl, I think that's because python and Ruby are 'cool' languages, and MS is trying to be like someone's Dad, 'getting hip with the kids'. I doubt it'll ever create an IronPerl simply because there's no mileage in it for them to entice the Perl developers over to Windows unlike the Python and Ruby folks that they're scared of losing to other platforms.
and all they had to do was post a made-up "home satellite" article to slashdot.
that's interesting.. I knew C# 3wasn't submitted to the standard, so LINQ and WPF aren't standards, but I didn't realise MS had added their own extensions to the standard already!
You don't happen to have some examples do you?
its not so much zealotry, if you code for Windows, you code using Microsoft libraries and everything that comes with Visual Studio. If it doesn't come with VS (or the platform SDK perhaps) then you don't use it.
I'm serious, things used to be a bit different, you could grab code from all over and use it without problem, but I've seen a definite shift away from that over the last few years.
Perhaps its because MS has supplied more library code, or perhaps its because of the Open source movement (ie managers have discovered people using free code, have read "free code is communist" on a website or magazine and have put 2 and 2 together to end up with "no use of free code in this organization").
So, nowadays, if you're going to write for Windows, you are very much pushed towards using the 'standard' or default libraries, and that means DirectX in this case.
OpenGL does have good libs, I like it, I've used it, but the number of Windows developers who'll use it is very small indeed. Its a shame and its why my attitude towards MS has shifted over the last few years, I am still a Windows developer, but I see the monoculture has grown too much.
More control means nothing, its all about the budgets.
If you have a budget, you buy as many windows licences as you need and you don't think twice about it (I'm being slightly simplistic of course). If your budget is cut, but you still must get the work done, you think "I wish I could get all those servers running without any licencing"... and you start to consider Linux.
There's a lot of people out there who "won't get fired for buying Microsoft", but not they cannot buy MS because they have no cash to spend, but they still have all that work still to do. What's a poor manager to do?
Then there's the technical route - you need to do something, you can't get a new VM installed (no Windows licence), so you install Linux and see... and perhaps you like it, and then you have 2 and 3 and pretty soon all your VMs apart from a couple of old legacy ones are running linux.
Maybe FOSS isn't cheaper overall, but it doesn't matter - its cheaper where it matters, in the upfront costs that the bean counters are counting. God knows there's enough inefficiency in IT (look at me, I'm writing design documents that pretty much simply rephrase everything in the requirement documents for no more than political reasons).
The answer is easy. Learn C++.
I'll get my coat....
J2EE v .NET is not the same comparison. Java v Net is.
Does it really scale better? Or is it that no-one notices any more because computers have got more RAM and faster CPUs nowadays.
the API is (well, was) almost entirely the old Java/J++ class library.
Now that said, you're still right. MS *is* better at development software than Sun. Visual Studio is best in its class and .NET is easier to use mainly because of that. I wouldn't like to type out those long, long namespaced methods without the IDE to autocomplete for me!
well, it snot that decent a little language. Its a MS port of Java with extra bits added mainly in the realm of the GUI and interop with legacy COM and Win32.
They have added extra features that are cool, like LINQ, but I feel they'll be heavily misused over the coming years. They've also adding features like extension functions which will make scripting languages look like statically strong-typed languages compared to C# as time goes by.
Its also pretty complex for a little language, I'm seeing a lot of people complaining on the web about how they're using up all their memory followed by little tutorials on Dispose and releasing objects so they actually get disconnected and then freed by the GC.
I think if it wasn't for curly braces, and the .NET class library, hardly anyone would be using it. Imagine if MS only released VB.NET, you wouldn't use it, the classic VB crowd wouldn't use it, but its practically the same language as C#.
BTW. C#2 is a standard, C#3 is not. MS has promised to submit it, but they havn't yet.
really? I thought MS has dropped support for OpenGL, apart from its very legacy v1.1 base. And you know they will never write a managed wrapper for it.
So, who will write a .NET game using opengl? If you're a Windows dev (and most game devs are) then you'll be using DirectX.NET, which oh so conveniently is not available for Linux.
So close. Yet so immensely far. Do you see the problem now? Its business a usual for MS, but with the added bonus of saying "But we are working on Linux interoperability, Mr DoJ, look - Mono".
perhaps you need to give away free copies of a Linux magazine with every netbook sold?
People who buy Windows often don't know what they're doing with them anyway, give eg Linux mag away, and you'll get a lot less returns and happier customers.
on thew other hand, itr would be acceptable to sign a 'never compete ever ever again' contract, if they were prepared to pay you for the duration of that contract.
Most contracts that do this have limited non-compete clauss, like selling a shop, you promise not to open another one in the same town for 6 months to a year. After that though, you can do what you like. Such terms are written into the contract, Id say they were unenforceable otherwise.
If we're lucky he'll add a few extensions to Mono :-)
(yeah, that assumes Mono will become more popular than C# :( )
for the same reason MS can't stop the word "Windows" (or "Word" for that matter) from being used by other entities.
Really? That won't stop them trying.
Open source means the source is publically available.
and the Windows source code itself, if you sign all the legal NDAs in their 'shared source' program (and they let you, of course). So I guess you think Windows is an Open source program?
Nice. You've just destroyed the common-usage of the term. You'll be telling us 'gay' means happy and has no homosexual meaning next?
Words change their meaning over time and with majority usage. "Open source" nowadays refers to a particular system of free and shareable code. That's just the way its gone, you can try and change it, but the majority of people have to agree and use it in your terms for it to actually change like you want.
only to your mind, not to that of a business executive who feels the need to protect his company's intellectual property and maximise further corporate innovation through aggressively pre-empting market share downward factors to further enhance shareholder value. (and his bonus).
what you or I think should happen is nothing when it comes to making money.
I refer you to the Asus ExpressGate technology - an embedded linux system on their newer mobos.
awesome porn... starring the poster, obviously.
C# v3 is not a standard.
From TFA:
Do you expect C#3.0 to become an ECMA and ISO standard, as previous versions have?
We're certainly open to that. There's no ongoing work in the standards committee at the moment, but it's really more a question of whether the community of industry partners out there would like to continue with that process. I should also say that the standards for C# explicitly do permit implementers to have extensions to the language, so though C# 3.0 is not standardized, it is certainly a complete implementation of the C# 2.0 standard. It is 100% backwards compatible, as all versions are.
no, if MS said No to Mono, those guys would bitch (quietly) about it and then start blogging about how dangerous it was to use an unsupported platform on an unstable operating system that did not have the quality integration of a well-tested stack of applications that only Microsoft can provide.
They'll do it to save you from yourself.
MS wouldn't care.
(from TFA: "It is possible to build alternate implementations. We are not building .NET for Linux, because the value proposition that we can deliver to our customers is a complete unified and thoroughly tested package, from the OS framework to databases to Web servers etc.")
actually, Microsoft called, said it'd like to go back to 2001 to try to take over the world some more:
On the failure to realise early enough the importance of internet search, Mr Ballmer said: "Do I wish we'd started the investment in search a few years earlier? Yes."
"We may be the David up against Goliath but we're working on it."
And he added that the real concern was the lead that Google had built up in online advertising.
"We probably missed the power of the advertising model, not so much the technology," Mr Ballmer said.
See, things have changed, once they'd have claimed it was the technology, honest. Now they don't give a fig about the tech - just the cash.
Actually, they did fix the problem... since it was an issue with their code and not C#. C# was doing what it was supposed to.
and in related news, C++ is not hard, its just that you're not intelligent enough to use it. And javascript is not slow, its just that you're using it for applications with incorrect performance requirements, and java isn't memory-hungry, you're just running it in a constrained environment :-)
If you say that C# is working fine, its a problem with the coders, then my points above are equally true.
I've been looking through stackoverflow recently, and there are a lot of questions along the line of "my .net app uses too much memory", its one of the inherent problems of a GC architecture that you need to be aware of, just like C#'s other language features. I believe its a big issue that needs more visibility - as you say, its not the GC that's at fault, its the whole understanding of how to properly code in a GC environment.
Unfortunately, most .NET coders are drawn to the language because it has features like this that are designed to make it easier for them to code, so they don't usually want to know that it is harder than they expected.
apart from the dodgy replication (new '911' ambulance customer, 2 sites, SQL 2005, clustered DB... support engineer called out at 2am because 1 DB had fallen over and the other one couldn't be bothered to take over)
Microsoft later shipped us a patch, which was nice of them, its just a pity that the latest rollup (v6) didn't contain the patches contained in patch v5.
Usually SQL Server problems are down to its locking strategy. Anyone coming from an Oracle background to work with SQL Server can easily get a totally unscalable DB. Our old product had to be extensively redesigned to work with SQLServer because of this 1 issue. 2008 apparently supports row-level locking much better so perhaps that's what he means by "getting it right" at last.
isn't that an indictment of their legacy development practices. If I look at all the projects I have here, I reckon I could consolidate a dozen into 1 very easily. I might, one day, get to merge the 3 almost-identical extract applications into just 1.
Now his app runs in 2 hours, if he'd refactored all that C++ code into a single C++ app, it'd probably run in 10 minutes :-)