C++ for stable future?

[Oscar]

C++ replaced quicker than COBOL

But if we compare with COBOL, is it resonable to think C/C++ will be replaced quicker? (I can imagine that COBOL is good at what it is doing, and C/C++ had some disadvantages compared to it)

Oscar

[VectraMan]

In my experience maintainability is due to design and organisation rather than choice of language.

I agree. It's just as easy to make a complete mess in any language.

Having said that, I do think there are a lot of positive things about Java and C# in as much as being a tidied up C++ with 99% of the power. But at the end of the day if you don't want limitations and want full control, C++ is the best tool.

[BlasterBates]

The performance difference at runtime between java and C++ today is negligible. Sounds like the problems there are more architecture rather than down to choice of language.

The difference in time taken to create and maintain programs in Java and C++ however is not negligible.

Maybe Java has improved but a few years ago, the bank I worked at had a lot of problems with things like garbage collection (completely stopping the application at peak trading times), incompatible java libraries etc etc, to the extent where the applications were no easier to maintain. In any case the main problem with maintenance in C++ was due to lack of standard containers (eg vectors), a major headache which has now been solved.
Developing in C++ these days is quite simple.

In my experience maintainability is due to design and organisation rather than choice of language.

[DodgyAgent]

I thought you were going to start being nice?

It was too good an opportunity to miss

[BrilloPad]

Have they dumbed the requirements down enough to get Churchill out of his permie job?

I thought you were going to start being nice?

[DodgyAgent]

I have recently gotten more calls for C++ positions. Seems they ask less experience than a few years ago. So I am thinking a bit that if I strengthen my experience here, I would be sure about stable well paid work for many years, without tracking moving targets like .NET or Java. What do you think?
The situation is a bit parallel to COBOL. What can we learn from COBOL? (Rates etc.)

Oscar

Have they dumbed the requirements down enough to get Churchill out of his permie job?

[bobhope]

The performance difference at runtime between java and C++ today is negligible. Sounds like the problems there are more architecture rather than down to choice of language.

The difference in time taken to create and maintain programs in Java and C++ however is not negligible.

[BlasterBates]

But I think you've just highlighted the problem "the servers struggle to keep up" , if this is a key risk management system then it is a handicap doing it in Java, and the other banks take the business. Deutsche Bank nicked a lot of FX business from other banks because of super fast messaging system. Admittedly this was the pricing and execution but the risk systems also need to provide realtime exposure.

[BrilloPad]

I would add I work on risk management on C++ and in my opinion because of performance it is unthinkable to use Java or .NET. There are two big risk systems here, both C++ but with Java front ends.

I also work on Risk for an IB - servers on Java with c++ front end - but there is a proposal to move the c++ to Java. The current system struggles to keep up. But I think the management want to move all the developers to India.

[BlasterBates]

I would add I work on risk management on C++ and in my opinion because of performance it is unthinkable to use Java or .NET. There are two big risk systems here, both C++ but with Java front ends.

[thunderlizard]

The big difference is that all the colleges and universities taught C++ up to about 2000, when Java and later C# took over. Yes there's been a bit of a fall in demand since then, but the sudden drop in supply outweighs it.

[VectraMan]

I hope so. And there's more to the industry than putting front ends on databases (which is what most of you seem to be doing).

I'm still doing MFC. I'd rather not be doing that at all to be honest, but it seems there are people with MFC apps and that part of my experience is just as important to them as general C++. However I'd probably be earning more if I'd spent the last 5 years putting front ends onto databases. I'd have probably have killed myself through boredem, but at least I'd be able to afford a better coffin.

[kingshuk]

From what I have seen in my current place of work (Investment bank)-

There are a fairly large number of C++ (on Unix) apps that has to be maintained (bug-fix, enhancements etc). C++ developers are required for this. It is gradually becoming difficult to get good C++ developers as some (many?) old C++ hands migrate to java/.NET.

The current policy here is to write all new apps in Java. Front end is either web (preferably) or swing. It is acceptable to deploy vendor supplied .NET apps. But no inhouse development in .NET (yet).

There are a few specialist areas (mostly to do with heavy numerical calculations) where new C++ programs are being developed.

Other than the injunction on .NET/C# I know that many City based organization has a similar policy/situation.

Outside the City many large organizations - public and private sector are preferring packaged apps (Oracle, SAP etc) over inhouse (or outsourced) custom development. So lot of work in Oracle and other packaged solutions are coming up.

So to sum up -

1. IMO more and more new developments will happen in Java, C#/.NET and packaged apps.

2. Due to so many C++ apps developed in 80s/90s - the demand for C++ developers will not die any time soon. But it will reduce. As developers in this become scarce the rates _MAY_ rise too.

3. Some specialist sectors (telco, financial risk modelling) will continue to use C++ due to existing large investment and because it makes sense to use C++

[BrilloPad]

I reckon c++ stuff will be less likely to go to India.

[Oscar]

I have recently gotten more calls for C++ positions. Seems they ask less experience than a few years ago. So I am thinking a bit that if I strengthen my experience here, I would be sure about stable well paid work for many years, without tracking moving targets like .NET or Java. What do you think?
The situation is a bit parallel to COBOL. What can we learn from COBOL? (Rates etc.)

Oscar