RE: [cinjug-users] Java vs. C

["Hudson, Loren (GE Infra, Aviation, Non-GE, US)" <loren.hudson@xxxxxx>]

Ed,

 

I'm not using JNI.  I pretty much start with the C code, rename it the file to .java, and work from there.  I'd say I end up with about half my Java code and half untouched C code.  Most of the C code is parsing through a very large array of unsigned shorts.  Since the syntax is the same WRT array subscripts, bit-shifting, and bit-masking, I'm leaving most of that code untouched.  I haven't even changed variable names at this point.

 

The I/O and variable declarations are really the only part I have to port.  That wraps the (largely) untouched array-parsing code. 

 

If I'm still worrying about this code a month an a half from now, I will be very concerned.  All told, it's less than 1000 lines of C code. 

 

 

To clarify, I was just wondering if there was any C syntax that compiles perfectly as Java syntax but doesn't do the same thing.  For instance, does C deal with floating point variables in the same way?  I've changed all of the floats in C to doubles in Java.  Nothing goes past the fifth decimal point, so could that be an issue?

 

James,

The C code reads in the data as unsigned shorts.  When it has a value that takes up more than two bytes, it concatenates them together when it outputs them.  None of the aggregated data comes from fields that are more than two bytes.

 

Josh,

Rounding may well be the culprit.  I'll look into that.  Thanks!

 

Thanks all,

 

Mike Hudson
MDW - Military Data Warehouse
Java 
Developer/Solution Architect
Sogeti Consultant
loren.hudson@xxxxxx

 

---

From: esumerfd@xxxxxxxxx [mailto:esumerfd@xxxxxxxxx] On Behalf Of Edward Sumerfield
Sent: Monday, October 15, 2007 2:43 PM
To: users@xxxxxxxxxx
Subject: Re: [cinjug-users] Java vs. C

That is a difficult question to answer.

You say you are 
wrapping the C code with java data structures. Does this mean that you are using 
JNI to call C methods?

How much of the C do you have to "port" as opposed 
to "wrap"? 

As a language C is as close to assembly as it gets so the 
biggest issues that you face are memory management. However, if you are not 
changing any of the original code then I don't see how your totals would be off. 

The key to these things is writing tests before each change so that 
after wards you know it failed. Then make changes in small increments so you 
find out what failed as soon as possible. Certainly easier said that done. 

My last port was a C++ to java and after a month and a half of trying to 
understand the original code and duplicate it in java I gave up, through the C++ 
away and spend the next 2 months with a customer re-writing it. I think this was 
much faster plus they were able to add features that were not in the original.

On 10/15/07, Hudson, Loren (GE Infra, Aviation, Non-GE, US) <loren.hudson@xxxxxx> wrote:

All,

 

I'm currently working on porting some C code over to Java.  The code opens a binary file, extracts the relevant data, and spits out a .csv file.  I'm getting some issues with the totals being off. 

 

The code was originally written by an engineer, not a programmer.  This is the fourth such program I've ported, and haven't had any real issues.  For the most part, I just wrap the already-written C code with some Java data structures, change the file i/o to use Java commands, and I'm done.  The program uses bit masking, bit shifting, and a whole host of global variables. 

 

My question is this:  Are there any pitfalls that you know of that I could be falling into with the C code that compiles as Java code?  I've never used C before, so this program is really the extent to which I know C. 

 

Thanks,

 

Mike Hudson
MDW - Military Data Warehouse
Java Developer/Solution Architect
Sogeti Consultant
loren.hudson@xxxxxx

-- 
Ed