Here's an interesting article about code generation from UML.

http://www.adtmag.com/article.asp?id=7850

I can attest to the huge productivity gains possible with these kinds of tools.  Over the last 12 years, I've seen several groups use schema driven code generators, and the coding methods to match.  I'd guess we get about 2x productivity improvements overall.

C# has features to help integrate with these tools.  Sather was also good in this reguard.

The two langauges that have the most code-generator related activity seem to be Java and C#.  It's worth watching.  It should be the next great thing in programming productivity improvements, but then again, it should have happened years ago.

Bill

Visual Basic, while not as expressive, powerful, or useful
as C or C++, does do some code generation in the form of
automatically creating subroutine and function calls to
match object events, methods, and properties. This somewhat
small thing allows very rapid application development. I
would like to see what I would call a 'modern' IDE for D
which employs the many timesaving 'tricks' of the more
popular commercial IDEs, but without going to the extent
that the IDE gets in the way of working directly with
code. For example, MS's MFC is what I consider a very bad
thing, in that perhaps of the programming that you do is
with 'wizards', which I dislike if they complicate or
obscure anything.

Bill Cox wrote:
> Here's an interesting article about code generation from UML.
> 
> http://www.adtmag.com/article.asp?id=7850
> 
> I can attest to the huge productivity gains possible with these kinds of tools.  Over the last 12 years, I've seen several groups use schema driven code generators, and the coding methods to match.  I'd guess we get about 2x productivity improvements overall.
> 
> C# has features to help integrate with these tools.  Sather was also good in this reguard.
> 
> The two langauges that have the most code-generator related activity seem to be Java and C#.  It's worth watching.  It should be the next great thing in programming productivity improvements, but then again, it should have happened years ago.
> 
> Bill
> 

In article <3F2A78DE.9090506@viasic.com>, Bill Cox says...
>
>Here's an interesting article about code generation from UML.
>
>http://www.adtmag.com/article.asp?id=7850
>
>I can attest to the huge productivity gains possible with these kinds of tools.  Over the last 12 years, I've seen several groups use schema driven code generators, and the coding methods to match.  I'd guess we get about 2x productivity improvements overall.
>
>C# has features to help integrate with these tools.  Sather was also good in this reguard.
>
>The two langauges that have the most code-generator related activity seem to be Java and C#.  It's worth watching.  It should be the next great thing in programming productivity improvements, but then again, it should have happened years ago.
>
>Bill

I seen a few good testimonials for code generation from modeling tools, including your stories.  Admittedly most of these seem to be in very specific domains of software development.  This makes sense because it's easier to tailor code generators for specific tasks.  The more general case seems to be harder.

I think the reason that Java and C# have a lot of activity is that they are easy to parse and have facilities for run time introspection. The makes the round trip from modeler to code back to modeler easier. C++ is more problematic.  I haven't run across a tool yet that could successfully reverse engineer any real life C++ code back into a model. Admittedly I've never been able to try any of the big name tools like Rational or Together soft. Their price tag is out of my league and frankly out of the league of most small companies.

However I am convinced that MDA in some form will eventually become the way the majority of software is developed.

For code generators check out http://www.softvelocity.com/

From their site
>Clarion is the foundation of the SoftVelocity product line and anchors >the company's reputation for fast, efficient database application >development. In addition to the Clarion 4GL language, the Clarion >product also includes both a C++ and Modula-2 compiler. All of the >languages share a common optimizer, and they can be mixed within a >single application.

Simon

Patrick Down wrote:
> In article <3F2A78DE.9090506@viasic.com>, Bill Cox says...
> 
>>Here's an interesting article about code generation from UML.
>>
>>http://www.adtmag.com/article.asp?id=7850
>>
>>I can attest to the huge productivity gains possible with these kinds of tools.  Over the last 12 years, I've seen several groups use schema driven code generators, and the coding methods to match.  I'd guess we get about 2x productivity improvements overall.
>>
>>C# has features to help integrate with these tools.  Sather was also good in this reguard.
>>
>>The two langauges that have the most code-generator related activity seem to be Java and C#.  It's worth watching.  It should be the next great thing in programming productivity improvements, but then again, it should have happened years ago.
>>
>>Bill
> 
> 
> I seen a few good testimonials for code generation from modeling tools, including your stories.  Admittedly most of these seem to be in very specific domains of software development.  This makes sense because it's easier to tailor code generators for specific tasks.  The more general case seems to be harder.  
> 
> I think the reason that Java and C# have a lot of activity is that
> they are easy to parse and have facilities for run time introspection.
> The makes the round trip from modeler to code back to modeler easier.
> C++ is more problematic.  I haven't run across a tool yet that could
> successfully reverse engineer any real life C++ code back into a model.
> Admittedly I've never been able to try any of the big name tools like
> Rational or Together soft. Their price tag is out of my league and frankly out of the league of most small companies.
> 
> However I am convinced that MDA in some form will eventually become the way the majority of software is developed.  
> 
> 
>