Stabilization

LE GARREC Vincent — Thu, 09 Nov 2017 22:13:28 -0000

Hi,

I just made a "big" commit about stabilization. Lot of wrong files creates lot of null dereference and other bugs.

Please, scream if I broke something. make check still working.

By the way, I will be happy if I could add some tests for ada and python. I never used them. For example, python looks to handle abstract method with library abc and annotation.

If someone may write the result for each test : just add python or ada in tests.sh, run make check and correct all .python.fail or .ada.fail. Please post the result, I will take in charge the modification in GenerateCodeAda.cpp and GenerateCodePython.cpp.

Regards.
Vincent

Dia2Code+ C++ generation

gilou — Thu, 20 Apr 2017 15:12:50 -0000

Hi,
Thanks for your quick reply.
Ok I was misled by the documentation : Here
In the examples section :

$ dia2code -t cpp -nc -d ~/C++ hw.dia
Will generate the .cpp and .h files for the classes in the test.dia diagram and put them in the directory ~/C++. It won't overwrite any existant file.

Regards,
gilou

Dia2Code+ C++ generation

LE GARREC Vincent — Thu, 20 Apr 2017 15:03:04 -0000

Hi,

dia2code only generates header. It doesn't generate prototype in source file.

Regards,

Dia2Code+ C++ generation

gilou — Wed, 19 Apr 2017 16:53:45 -0000

Hello all,
Thank you for your work, I'm happy to use this new version plugin. I'm learning UML for C++ project and it's simple with dia. So my question is : I've draw a class diagram and I try to generate .cpp and .hpp files but dia2code+ give me only .hpp files.

I've tried the command : dia2code -t cpp -d C++/ diagram.dia

.hpp files generation are already a great help.
That come probably from my diagram but I don't know why. I've tryed the example1.dia from dia's example and it's the same.
I can send you my diagram if needed.

Regards,
gilou

Refactoring dia2code

LE GARREC Vincent — Fri, 06 Mar 2015 18:03:47 -0000

Maybe I spoked too fast. I think that both ways to implement interface should be keep.
Please find enclosed an example where two classes implement the same Interface. With Classe2, the class Interface will be needed even if no direct connection are made.

Refactoring dia2code

LE GARREC Vincent — Wed, 04 Mar 2015 11:35:37 -0000

I'm thinking about implement interface. Dia2code 0.8 have it so it's necessary that the new one have it too.
Actually, Realization and Generalization are doing the same thing: inheritance. Implements are doing an other (lolipopImplementation function. I don't have tested it and I don't really know what it's doing).

I would like to change it:
Realization will implement an interface (implements in java). If a class in pointed by a realization link, it will automatically be converted from class to interface,
Generalization will create an inheritance (extends in java),
Implement (with function lolipopImplementation): don't know what it's doing. Remove this feature.
Source: http://www.informit.com/articles/article.aspx?p=29224&seqNum=2 and Fig 4 and 5.

Refactoring dia2code

Oliver Kellogg — Thu, 19 Feb 2015 18:15:37 -0000

I'm trying to add const stereotype for Java but I have a problem:
Actually, in C++ generator the result is:
const type constance = 150;
constance is the name of the "transformed class", type is the Type
of the first attribute and 150 is the Value of the first attribute.
The name of the first attribute is ignored and the other attributes
are also ignored.

Hmmm... I think there is no straight forward answer because the programming languages are so different with respect to constants.

The question is whether we want to support package level constants.
Doing that would create a problem for Java because there is no such thing as a package level constant in Java.

Therefore I would suggest that we only support constants within classes.
Then, applying the «const» or «constant» to an attribute would produce a constant within the class. In this case, the attribute name becomes the constant name and the attribute default value becomes the constant value.

In C, one would probably ignore the class name.
(IMHO, doing UML with C language target is not a very good match to start with...)

Refactoring dia2code

LE GARREC Vincent — Wed, 18 Feb 2015 22:32:23 -0000

Done. But the change of name is only effective with -1 argument (one class one header, so by default for java).

If you have other idea, you're welcome.

PS: if you have an opinion about https://sourceforge.net/p/dia2code/discussion/48621/thread/55447541/?page=1#0858/2bd5/7ba0 …

Refactoring dia2code

Oliver Kellogg — Sun, 15 Feb 2015 23:02:03 -0000

About changing convention of filename, are you sure that's a good idea?

Actually, the idea comes form the GCC Ada compiler (GNAT), more precisely from its file naming convention for child units.

This will create long name no?

In practise, I haven't felt this to be a problem.
Of course, if you have a_long_package_name with an_even_longer_nested_package_name containing an_embedded_class, then you will have
a_long_package_name-an_even_longer_nested_package_name-an_embedded_class.hpp
instead of just an_embedded_class.hpp.
The advantage is that the file name directly reflects the package names.
Which could be an advantage when dealing with same named classes in different packages.

Refactoring dia2code

LE GARREC Vincent — Sun, 15 Feb 2015 14:09:41 -0000

Done. I finished the C# generator.

About changing convention of filename, are you sure that's a good idea? This will create long name no?

Recent posts to Discussion

Stabilization

Dia2Code+ C++ generation

Dia2Code+ C++ generation

Dia2Code+ C++ generation

Refactoring dia2code

Refactoring dia2code

Refactoring dia2code

Refactoring dia2code

Refactoring dia2code

Refactoring dia2code