non-OO -> UML (was: Re: (FC-Devel) Hi)

[Matthias Pfisterer <Matthias.Pfisterer@gmx.de>]

Hi Roman,

a friend of me is doing work on reverse engineering non-OO code to UML
models as a master's thesis. Here is his email:
mailto:schielgr@rupert.informatik.uni-stuttgart.de

Matthias


> Roman Razilov wrote:
> 
> Hi,
> 
> This list seems to be not very active, but still I hope on some
> reaction.
> 
> May be this mail is offtopic, but I'm quite a newbee in CASE et all.
> actually, I've no regular education as IT professional.
> 
> I've being hackin for 10+years, and figured out for me that the most
> time I don't envent anything, the most time I had just to take one
> pice of existing source code and built in in to another programm, or
> rewrite it for using with another library, on another language for
> another OS. It makes me kind of sick.
> 
> There are grate ( and less grate ) packages to do OOA, OOD, OOP but
> most of them are just braindead charting tools with code-generating
> features.
> 
> The problems they have is that no way to use all that code that one
> allready have.
> 
>  90% all existing programms, especially common business application,
> are more than 10 years old and have nothing to do with  OO, (COBOL
> never die ) look at ol this ERPs..
> 
> There is a lot of programms written to use not OO ( or semi -OO ) GUI
> libraries ( all that Motiff, gtk++ etc. stuff ), they are greate, but
> to port it to newer GUI is often simple but dumb work.
> 
> I don't belive that just another charting tool with even better
> intergration one specific platform cann chage the situation.
> 
> I think, that the Real Good Thing is a suite of following 5 basic
> tools
> 
> 1. Reverse-engenering tool ( Fuzzy, also that can copy with not OO
> styled, even buggy code )
> 2. Model simplifier ( The real program are rare designed from scratch,
> almoust usual is 5 year old code with some messy hacks did by
> not-expirienced programmer) It must expand the code to some standard
> (three-tire etc.)
> 
> 3. Some graphical manupulation tool to alter model, with possibility
> _dynamicly_ analyse the processes. ( A kind of interpreter of UML )
> It's quite nessesary to understand how it works, anlalyse concurrency
> and performance issues..
> 
> 4. Solution database ( Tool to extract and recognise algorythmen )
> i.e. if I have buble sort integrated somethere in code, I expect that
> system recognize it as sorting algorythm, and that I easely can
> replace it with other more.
> 
> 5. Code generator ( even optimising / reducing ) if I have software to
> 100% generated from model, it is a Good Idea to make Hi-Level
> Optimisation, throw away not used methods, even generate some pure C,
> PASCAL or even Assembler code ( UML+-> ASM compiler ;-)
> 
> This is not to do software develpment but re-development. Recently
> I've lloked through job-adds, and I've seen  that although there is a
> lot of demand for programmer, the jobs are mostly not exciting.
> Customising and re-develop...
> 
> Maybe it's crasy, but I dream about it.
> 
> I'm looking for someone to discuss, or someone who can tell me "forget
> it, company XYsoft sells it allready for gazillion $".
> 
> 

-- 
Matthias Pfisterer	<mailto:Matthias.Pfisterer@gmx.de>

        Share your knowledge.
        It's a way to achieve immortality.

                 (from a nepalese mantra)

Java Sound Examples:
http://rupert.informatik.uni-stuttgart.de/~pfistere/jsexamples/
Tritonus, the open source implementation of the Java Sound API:
http://tritonus.sourceforge.net/
--------------------------------------------------------------