[download2live]

Originally Posted by Samurai

T
After doing STL for 15 years, then I delved into boost c++.

Fancy writing a book on it?
There official documentation is really well, terse.

And the fact remains that the C++ market is dwindling to sectors where performance matters.

The Hardware is getting cheaper and Programmers are getting more and more expensive.

[Samurai]

Originally Posted by download2live

Fancy writing a book on it?
There official documentation is really well, terse.

There is already a very good book on it. Just $6 on Kindle. http://www.amazon.com/The-Boost-Libr.../dp/0982219199

Originally Posted by download2live

And the fact remains that the C++ market is dwindling to sectors where performance matters.

Still in the top 5 after 30 years.

Originally Posted by download2live

The Hardware is getting cheaper and Programmers are getting more and more expensive.

Been hearing that since 15+ years myself. But performance never goes out of fashion. But it is the design flexibility that keeps me with C++, one can do magic with it thanks to Boost. It doesn't feel like work at all.

[download2live]

Originally Posted by Samurai

There is already a very good book on it. Just $6 on Kindle. http://www.amazon.com/The-Boost-Libr.../dp/0982219199

Still in the top 5 after 30 years.

Attachment 1121956

Been hearing that since 15+ years myself. But performance never goes out of fashion. But it is the design flexibility that keeps me with C++, one can do magic with it thanks to Boost. It doesn't feel like work at all.

Not really. I never go with that fancy graph. The ground realities are different. C++ might still used in Product Companies but there also it is getting relegated to areas where performance matters, like Automated trading systems core engines. It is painful to learn and hence must be good mantra somehow is not valid anymore. C# is nothing but C++, with lot of design patterns already implemented and ready to use.

In fact I feel the success of Java and later on .NET is significantly due to the royal pain inflicted by these languages like C++. Boost being a case study. Awesome as it is, the official documentation still leaves a lot to be desired.

Not taking anything away from C++. It is awesome. But I see it losing ground.

[sgiitk]

Let me put in a joker - even today more than 90% high end Scientific Computing is in (hold your breath) FORmula TRANslator aka FORTRAN. So should we demand FORTRAN skills in our postgrads! Once you have a good grounding in structured programming, switching is not such a pain. Im my career I have worked in FORTRAN, Basic, Pascal, Modula, C, as well is the Graphical Language LabVIEW.

[Arch-Angel]

that is true, sg sir. the reason being most of the legacy systems built during the last few decades are all mostly complete systems built with fortran by itself. i had heard from someone that the number of systems running with fortran and the number of 'lines of code' is far greater than all of the programming languages combined.

with regards to c++, i think it will continue to live on but i personally love the versatility of c# and how .net framework handles memory management and obviously, the jit compiler and the garbage collection routine which makes life slightly easier. also, i used to love c++ and its way of dma, malloc,calloc,realloc. the funniest thing i have seen is people using the wrong operator for cin and cout and complaining of not getting anything .

[Samurai]

Originally Posted by download2live

Not really. I never go with that fancy graph. The ground realities are different.

Just because you call something fancy, it doesn't become wrong. Tiobe index is the most respected report on this matter. If you want to discredit a graph, you should look at how the graph was generated. Here is the August report, they have enhanced it further by adding more search engines. Read the index definition to understand how it is done, now C++ is more popular on the web than Objective-C. Even I am finding it hard to believe, but we can't deny the facts.



Still this doesn't reflect the full ground reality. Older languages don't get discussed much on the web, because people have very few questions. More on that later.

Originally Posted by download2live

It is painful to learn and hence must be good mantra somehow is not valid anymore.

If you are talking about auto-transmission vs manual transmission, you are right. But you can't apply that rule to everything. Quantum theory is very difficult to learn, does it make it bad or not worthwhile? Something things are tough, calling it sour grapes is not the right way.

Originally Posted by download2live

C# is nothing but C++, with lot of design patterns already implemented and ready to use.

I did learn C# a while back. For a seasoned C++ programmer it is like downgrading from superbike to a moped. So I ditched it. We still use it for web applications, although we are moving all the webs to python. But in my company anybody with 3+ years experience moves over to serious stuff using C++.

Originally Posted by download2live

In fact I feel the success of Java and later on .NET is significantly due to the royal pain inflicted by these languages like C++.

Absolutely true, I have seen it happening since Java was created. It is very hard to create good C++ programmers, you need very technical minds for that. But much easier to create Java/C# programmers.

Originally Posted by download2live

Boost being a case study. Awesome as it is, the official documentation still leaves a lot to be desired.

That is true, I had to hunt a lot get better documentation. Finally I found the right book to crack the mystery. Check the link I provided.

Originally Posted by sgiitk

Let me put in a joker - even today more than 90% high end Scientific Computing is in (hold your breath) FORmula TRANslator aka FORTRAN.

That is very true. But scientific computing is now a small part of total computing world.

Originally Posted by Arch-Angel

that is true, sg sir. the reason being most of the legacy systems built during the last few decades are all mostly complete systems built with fortran by itself. i had heard from someone that the number of systems running with fortran and the number of 'lines of code' is far greater than all of the programming languages combined.

Actually no. There is a bigger elephant in the room. The biggest codebase belonged to COBOL, followed by FORTRAN. But you will never see these people talking about their languages excitedly on the web. However, the last 20 years has changed the demographics, thanks to explosion of Intel based PC/Servers running Windows/Linux, smartphones and tablets. We might have 20 times more programmers compared to the 80s, who don't do COBOL/FORTRAN. I would love to see the latest demo-graph of codebase, but I also know it is impossible to collect that data.

Originally Posted by Arch-Angel

with regards to c++, i think it will continue to live on but i personally love the versatility of c# and how .net framework handles memory management and obviously, the jit compiler and the garbage collection routine which makes life slightly easier. also, i used to love c++ and its way of dma, malloc,calloc,realloc. the funniest thing i have seen is people using the wrong operator for cin and cout and complaining of not getting anything .

Do you mean C? functions like malloc,calloc,realloc are used in C, not in C++. Good C++ programmers know how to manage memory (more so with shared_ptr), it becomes second nature.

[DerAlte]

Originally Posted by download2live

... C++. It is awesome. But I see it losing ground.

That's a strange way of putting it. I find it amusing that languages are compared - more so with what can at best be called 'environments' (.NET).
For example,

Originally Posted by sgiitk

... more than 90% high end Scientific Computing is in (hold your breath) FORmula TRANslator aka FORTRAN. ...

Ditto for COBOL, which hasn't lost it's pre-eminence in large enterprise applications running on Mainframes. And to think both languages are 50+ years old!

Generation after generation, 'languages' (I use quotes, since there are many primary language interfaces which are graphical, like MatLab, MatriXx, LABView etc. which ultimately produce executable code via libraries written in C) have evolved or been invented for specific purposes like FORTRAN and COBOL. C++ came about to prevent large team programming efforts in C language suffering from 'domain' transgressions (one function messing up the data of another function, possibly written by a person in another country).

PERL, Python, CURL, PHP, even Java - all were 'invented' to solve specific problems, but terming any of them as the best language would be stretching it. One can use a language as a substitute for the most suited one (language fan-following surpasses religious fanaticism sometimes), but that is always like trying to fit a square peg into a round hole. If a target environment does not *ever* need to be ported, saying Java is the best language instead of C/C++ is like looking at the animal from the wrong end.

Originally Posted by sgiitk

... Once you have a good grounding in structured programming, switching is not such a pain. ...

Not sure whether you have noticed, 'programmers' today expect design to be created by a 'designer' (or the Team Leader or the PM - whoever is the closest to lay blame on), and be provided building blocks (not different from Lego blocks used by children) with which they will put together a part of an application 'as per design provided by designer'.

Very few of them have a clear understanding how an application - whose creation they are participating in, in however small a way - is supposed to work for the USER. Almost all programmers understand the application from inside out, i.e. extrapolating and projecting what the application is doing based on what code they are involved in. Very similar to the 'Five blind men and an elephant' story. So if the pain in life (writing a program usually is that for most) for a programmer is writing the Administration functions, he will explain that the User first opens the Administration screen and sets up further users etc. etc. - every time the application runs.

Perceptions of programming, and hence expectations and habits while working, have been irreparably damaged once fill-in-the-blanks programming environments (like Oracle Developer2000, Visual Basic etc.) made an appearance. Those who were once excellent C/C++ programmers, suddenly switched to Dev2K and VB jobs - because companies were paying more for those jobs. Many of those made a 2nd switch to becoming SAP ABAP programmers, again because the pay was higher! Such expectancies were bolstered by C#/.NET, even worse - VB/.NET, resulting in customers being disillusioned with output from most such individuals / teams.

[carboy]

Originally Posted by Samurai

There is a reason for that, then Borland C++ was the only compiler that supported templates. Guess in which compiler STL was developed? Yup, Borland C++. So I had started using Borland's template library even before STL was created.

Are you sure about this? Stephanov developed the STL when he worked at SGI. I would assume he would have used an IRIX C++ compiler for this.

I think Watcom compilers had templates even before Borland but I am not sure. But yeah, Borland did a lot of important work in templates. But most compilers support the STL as it originally existed for more than 15 years now.

For me the biggest shock was when I moved from Stroustrup's C++PL 2nd edition to 3rd edition. I understood the 2nd edition reasonably well (at least I thought I did) - it was pre-STL. And I didn't understand the first chapter in the 3rd edition - that prompted me to learn the STL (back in 1997, I think). Before the STL, the Rougewave library used to be somewhat common for containers and algorithms.

I worked on C & C++ for most of my life. But late in life, I started doing Java and I absolutely love it - you are far more productive in Java than in C++. Joel says the same thing in one of his articles - the most productive programmers are C++ programmers programming in Java.

[Samurai]

Originally Posted by DerAlte

Not sure whether you have noticed, 'programmers' today expect design to be created by a 'designer' (or the Team Leader or the PM - whoever is the closest to lay blame on), and be provided building blocks (not different from Lego blocks used by children) with which they will put together a part of an application 'as per design provided by designer'.

Thanks for bringing up a very good point. C++ programmer has to be designer. If not, he is not doing C++ programming at all.

Originally Posted by DerAlte

Perceptions of programming, and hence expectations and habits while working, have been irreparably damaged once fill-in-the-blanks programming environments (like Oracle Developer2000, Visual Basic etc.) made an appearance.

I once used PowerBuilder in 1993. I felt like a data entry operator.

Originally Posted by carboy

Are you sure about this? Stephanov developed the STL when he worked at SGI. I would assume he would have used an IRIX C++ compiler for this.

No. Alexander Stepanov & Meng Lee worked in HP when C++ STL was unveiled to the world. I clearly remember BC++ because I was using it then. Borland had a very nice template library with many containers.

Originally Posted by carboy

For me the biggest shock was when I moved from Stroustrup's C++PL 2nd edition to 3rd edition. I understood the 2nd edition reasonably well (at least I thought I did) - it was pre-STL. And I didn't understand the first chapter in the 3rd edition - that prompted me to learn the STL (back in 1997, I think).

I remember, I was quite puzzled why that book started with STL.

Originally Posted by carboy

Before the STL, the Rougewave library used to be somewhat common for containers and algorithms.

Yes, I too used Roguewave. I had Tools.h++ professional which included very nice internet library for http/ftp/smtp protocols. Then in 1999 they went crazy and started demanding runtime royalty. That is when I had to ditch them.

Originally Posted by carboy

I worked on C & C++ for most of life. But late in life, I started doing Java and I absolutely love it - you are far more productive in Java than in C++. Joel says the same thing in one of his articles - the most productive programmers are C++ programmers programming in Java.

I used Java for 3 years and then stopped using it. It was actually messing around with my design philosophy. Been 10 years since I wrote Java code. I generally agree with Joel's thoughts, but not always. Have you checked Boost, it supercharges C++. It opens up a whole new level that is unthinkable in Java.

[carboy]

Originally Posted by Samurai

No. Alexander Stepanov & Meng Lee worked in HP when C++ STL was unveiled to the world. I clearly remember BC++ because I was using it then. Borland had a very nice template library with many containers.

You are right. My memory is screwing with me.

By the way, Stephenov wasn't a fan of OOP - he thought it was a hoax.

Originally Posted by Samurai

Have you checked Boost, it supercharges C++. It opens up a whole new level that is unthinkable in Java.

Yeah, I used boost in 2010, IIRC. And I think a lot of boost stuff has been added to the C++11 standard with which I am woefully out of touch. However, I am not programming much these days so no incentive to go deep into it.

[DerAlte]

Originally Posted by Samurai

... I once used PowerBuilder in 1993. I felt like a data entry operator ...

I wrote an ERP in 1999 for SME discrete manufacturing industries virtually alone (with my wife's cousin who had just finished Engg then) with PB and Sybase - just for the heck of it.

It is still running in many places in Bangalore. My wife's cousin maintained the installations initially till he joined IBM. After that his wife has been maintaining it.

The advantage of PB was simplicity and robustness. It is still very popular, considering IBM has a division which does only PB-based maintenance.

[Arch-Angel]

Ah samurai sir, yes I think it was through cstdlib.h header. Native use of pointers and malloc and calloc were in c, anyways c++ is c with classes .
How is the market regarding SharePoint these days?

[carboy]

Originally Posted by Arch-Angel

Ah samurai sir, yes I think it was through cstdlib.h header.

There is no <cstdlib.h>. There is <cstdlib> in C++ and <stdlib.h> in C.

Originally Posted by Arch-Angel

Native use of pointers and malloc and calloc were in c, anyways c++ is c with classes .

You are trolling, aren't you?

[Arch-Angel]

Originally Posted by carboy

There is no <cstdlib.h>. There is <cstdlib> in C++ and <stdlib.h> in C.

force of habit and i ended up with a .h. absolutely right. it is cstdlib on c++ and stdlib.h on c. and if you thought i was trolling with regards to c++ being c with classes, that is how we taught in our college days lol. but how is the market currently with regards to sharepoint? i would love to hear an answer for this. i could go on and on how much less an emphasis was given during our school/college days, but its better to learn than sit and crib about what was not done. i would love a small writeup by any of knowledgeable folks about what is expected from a person in a development/programmer role.

[DerAlte]

Originally Posted by Arch-Angel

... How is the market regarding SharePoint these days?

Do you mean how is Sharepoint selling? Or what opportunities are there for programmers to work on Sharepoint-based applications?