dia2code-general — Dia2Code general discussion

[Dia2code-general] Enumeration

[Carsten Z. <car...@at...>]

Hello All,

I stumbled over dia2code today and am pretty impressed. Usually I code
my idl and java files by hand after designing in dia ...

But using dia2codes leaves to questions:

1. When generating java, how should the diagram look to generate java
interfaces, now I get abstract classes ?

2. Is it possible to model enumeration in a way that dia2code generates 
them as well in idl ?

Bye, Carsten

-- 
Carsten Zerbst <car...@gr...>
-- 
Carsten Zerbst <car...@at...>

[Dia2code-general] Enumeration

[Carsten Z. <car...@gr...>]

Hello All,

I stumbled over dia2code today and am pretty impressed. Usually I code
my idl and java files by hand after designing in dia ...

But using dia2codes leaves to questions:

1. When generating java, how should the diagram look to generate java
interfaces, now I get abstract classes ?

2. Is it possible to model enumeration in a way that dia2code generates 
them as well in idl ?

Bye, Carsten

-- 
Carsten Zerbst <car...@gr...>

[Dia2code-general] new release of dia2code?

[Richard T. <ric...@ht...>]

Hi all and especially Oliver,

when do you think it's tme for a new relase of dia2code?
Oliver are you planning any structural changes in the near future?


/Richard
-- 
Ph.D. Student
Dept. of Informatics and Mathematics
HTU

[Dia2code-general] New release of dia2code

[Richard T. <ric...@ht...>]

Hi all,

expect a new release of dia2code within a week or two.

That is to say, if the original developers don't show up.

We have changes in sql, idl and cpp.
We have added a c# generator.

More things that should be included?


/Richard
--=20
Ph.D. Student
Dept. of Informatics and Mathematics
University of Trollh=E4ttan/Uddevalla

[Dia2code-general] =?ISO-8859-1?Q?RE=3A=20=5BDia2code=2Dgeneral=5D=20Problem=20in=20Windows=20version=3F?=

[<tho...@ep...>]

-- Original Nachricht --

>Hallo everyone, 
>
>I am experiencing a problem which seems to be related only to the Window=
s
>version of Dia2code.
>Dia version: installed from dia-0.90-1-setup.exe
>dia2code:   installed from dia2code-0.8.1-setup.exe
>
>I created a small UML diagram containing 2 classes, and saved it as test=
.dia.
>dia2code resulted in the following output:
>
>C:\Program Files\dia\dia2code\bin>dia2code -t cpp test2.dia
>test2.dia:1: error: Start tag expected, '<' not found
>?=EF
>^
>That file does not exist or is not a Dia diagram
>C:\Program Files\dia\dia2code\bin>

you have to uncheck the "Save Compressed" Item in the Save-Dialogue.

-Thomas



________________________________________
Abos online bestellen. Oder Leser werben und Pr=E4mie aussuchen. http://w=
ww.epost.de/aboservice

[Dia2code-general] Problem in Windows version?

[Willem J.W. S. <wi...@di...>]

Hallo everyone,=20

I am experiencing a problem which seems to be related only to the =
Windows version of Dia2code.
Dia version: installed from dia-0.90-1-setup.exe
dia2code:   installed from dia2code-0.8.1-setup.exe

I created a small UML diagram containing 2 classes, and saved it as =
test.dia.
dia2code resulted in the following output:

C:\Program Files\dia\dia2code\bin>dia2code -t cpp test2.dia
test2.dia:1: error: Start tag expected, '<' not found
?=EF
^
That file does not exist or is not a Dia diagram
C:\Program Files\dia\dia2code\bin>

I get similar results when I select another language. I also moved =
everything to a shorter path name without spaces, namely: c:/temp, and =
the same results emerged.
When saving from Dia, I have tried both the Save As and the "Export" =
options to ensure that I export to the *.dia file format, but the =
results remain the same.
The file test.dia is certainly not corrupted, as it displays fine when I =
open it in Dia.

When I select the GUI menu in Windows Explorer to create a C file (or =
Java - Cpp is missing), then a dialog box flashes on-screen, but I =
cannot determine what is being displayed, before the display disappears. =
No output then seems to be generated.

Thereafter I tried it on Mandrake 9.0. I installed Dia from the Mandrake =
CD (version 0.90) and dia2code from dia2code-0.8.1-1.i686.rpm. I =
followed similar steps as in Windows, and the program created the =
desired output seemingly correctly.

Therefore, I presume the problem is either with the Windows version of =
Dia; or with the Windows version of dia2code. So, at this stage, it =
seems to me, the Windows version of Dia is broken.=20

Greetings
Willem Semmelink

[Dia2code-general] Contributors et.al.

[Richard T. <ric...@ht...>]

Hi,

I've added the patches to the cvs that I could find.

* SQL needs to be tested now (I can't do that)
* IDL has been touched - try it out.
* C# support is in - try it out.

Check out the new cvs and if you have fixes/enhancements etc you want to
contribute please send patches to the list or directly to me!



/Richard
--=20
Ph.D. Student
Dept. of Informatics and Mathematics
University of Trollh=E4ttan/Uddevalla

Re: [Dia2code-general] C# support

[Richard T. <ric...@ht...>]

I have added the tmp.tgz (patches @ sf.net) files changes into cvs.
(scan_tree.c and generate_code_idl.c fixes)
And I have added you to the AUTHOR text file.

If you have any more contributions please scream now! :)


/Richard



On Tue, 2002-12-03 at 21:13, Oliver Kellogg wrote:
> Hi there,
>=20
> Sorry for bothering you but I have been trying to contact
> Javier, and also Chris McGill who wrote the generate_code_idl.c,
> but without success.
>=20
> I have enhanced the generate_code_{idl,cpp,ada} to deal with
> nested packages, and wanted to contribute my mods back to the
> dia2code project. (For lack of response, I had made a preliminary
> patch to generate_code_idl.c which can be found in the "Patches"
> section of the Sourceforge dia2code project.)
>=20
> Nice to hear there are still others interested in dia2code.
>=20
> In case you do succeed in contacting someone, please let me know.
> Currently, I am not even sure the original dia2code authors are
> listening any more...
>=20
> Thanks,
>=20
> Oliver
>=20
> -- oke...@us...
> -- http://orbitada.sourceforge.net
>=20
> Richard Torkar wrote:
> >
> >
> > Hi Javier and others!
> >
> > I have created a generate_code_csharp.c which gives me basic
> > functionality for generating c# code. (I edited the makefiles and
> > main.c/dia2code.h/code_generators.h). And I have used it this week for =
a
> > project.
> >
> > It was very easy to do so by using Javier's generate_code_java.c ;)
> > Very nice code Javier, it is a pleasure to read it!
> >
> > What I've basically done is switching java keywords being used to c#
> > keywords. Now, I know that this is in no way complete _but_ it is a
> > start and by having it in cvs others might be tempted to help. I myself
> > am a Java developer so this whole new c# thingy is a bit new to me.
> >
> > Patch is attached.
> >
> > So tell me guys do you have any other suggestion as to how approach thi=
s
> > in a better way or do it the "right way"[TM] ;) As it is now it is more
> > or less the generate_code_java.c with a few minor tweaks.
> >
> > What should I do in which order so that we will one day have "complete"
> > c# support in dia2code?
> >
> >
> >
> > /Richard Torkar
> > --=20
> > Ph.D. Student
> > Dept. of Informatics and Mathematics
> > HTU
> >
--=20
Ph.D. Student
Dept. of Informatics and Mathematics
University of Trollh=E4ttan/Uddevalla

Re: [Dia2code-general] C# support

[Richard T. <ric...@ht...>]

Could you provide me with a link to your enhanced c generator?
I'll try to push it in cvs asap.

[Please make sure the patch applies cleanly against the cvs]

/Richard


On Tue, 2002-12-03 at 21:13, Oliver Kellogg wrote:
> Hi there,
>=20
> Sorry for bothering you but I have been trying to contact
> Javier, and also Chris McGill who wrote the generate_code_idl.c,
> but without success.
>=20
> I have enhanced the generate_code_{idl,cpp,ada} to deal with
> nested packages, and wanted to contribute my mods back to the
> dia2code project. (For lack of response, I had made a preliminary
> patch to generate_code_idl.c which can be found in the "Patches"
> section of the Sourceforge dia2code project.)
>=20
> Nice to hear there are still others interested in dia2code.
>=20
> In case you do succeed in contacting someone, please let me know.
> Currently, I am not even sure the original dia2code authors are
> listening any more...
>=20
> Thanks,
>=20
> Oliver
>=20
> -- oke...@us...
> -- http://orbitada.sourceforge.net
>=20
> Richard Torkar wrote:
> >
> >
> > Hi Javier and others!
> >
> > I have created a generate_code_csharp.c which gives me basic
> > functionality for generating c# code. (I edited the makefiles and
> > main.c/dia2code.h/code_generators.h). And I have used it this week for =
a
> > project.
> >
> > It was very easy to do so by using Javier's generate_code_java.c ;)
> > Very nice code Javier, it is a pleasure to read it!
> >
> > What I've basically done is switching java keywords being used to c#
> > keywords. Now, I know that this is in no way complete _but_ it is a
> > start and by having it in cvs others might be tempted to help. I myself
> > am a Java developer so this whole new c# thingy is a bit new to me.
> >
> > Patch is attached.
> >
> > So tell me guys do you have any other suggestion as to how approach thi=
s
> > in a better way or do it the "right way"[TM] ;) As it is now it is more
> > or less the generate_code_java.c with a few minor tweaks.
> >
> > What should I do in which order so that we will one day have "complete"
> > c# support in dia2code?
> >
> >
> >
> > /Richard Torkar
> > --=20
> > Ph.D. Student
> > Dept. of Informatics and Mathematics
> > HTU
> >
--=20
Ph.D. Student
Dept. of Informatics and Mathematics
University of Trollh=E4ttan/Uddevalla

[Dia2code-general] New version of dia2code or not?

[Richard T. <ric...@ht...>]

Hi,

send all patches (as I wrote earlier) to the list or me directly
(richard.torkar AT htu.se).

Make sure that the patches apply cleanly against dia2code cvs!

cvs -d:pserver:ano...@cv...:/cvsroot/dia2code
login

cvs -z3
-d:pserver:ano...@cv...:/cvsroot/dia2code co
dia2code
[watch out for line breaks in the above two lines).

Also make sure to send me short one- or two-liners about what this patch
really do!

I think Javier will forgive us for acting on this matter. After all this
is what Open Source is all about!

When we have stabilized the new dia2code (I'll test it), we can release
a new version (0.8.2).

After it has been released we need to have a discussion about the future
of dia2code. What features/languages/diagramtypes do we want to support
etc.

At that time (if for some reason the original maintainers have not
showed up) I will make sure to add more developers to this project (if I
can do that as a regular maintainer).


Regards,

/Richard
--=20
Ph.D. Student
Dept. of Informatics and Mathematics
HTU

RE: [Dia2code-general] C# support

[Richard T. <ric...@ht...>]

This sounds great (c#)!
I knew my solution was a bit rudimentary ;)

Ok, for everyone listening now on the list!
IF you have any patches you want to get into dia2code (C# sql idl, I
recall has been mentioned), send them to me directly (if it's big) or to
the list. (richard.torkar AT htu.se)

I have an admin account at dia2code which Javier gave me a long time
ago.

I'll see what I can do. I will release a dia2code 0.8.2 if the present
maintainers do not show up!


/Richard



On Wed, 2002-12-04 at 10:52, Thomas Hansen wrote:
> I did the exact same thing as you about two months ago, but the files wer=
e
> to big so they got stuck in the MailGroup's moderater's wait file or
> something...
> I think my is a bit more complete then yours, also it doesn't have the
> missing update "#DEFINE MAX_COMPILERS 9" (or what the define was), you
> should increase the number of compilers supported since the version you s=
ent
> will use uninitialized heap-memory I think...
> Also I used "///" as comments since this will work with the "Generate
> Comments Web Page" functionality in the IDE...
>=20
> -----Original Message-----
> From: Richard Torkar [mailto:ric...@ht...]
> Sent: 3. desember 2002 16:46
> To: dia...@li...
> Subject: [Dia2code-general] C# support
>=20
>=20
>=20
>=20
> Hi Javier and others!
>=20
> I have created a generate_code_csharp.c which gives me basic
> functionality for generating c# code. (I edited the makefiles and
> main.c/dia2code.h/code_generators.h). And I have used it this week for a
> project.
>=20
> It was very easy to do so by using Javier's generate_code_java.c ;)
> Very nice code Javier, it is a pleasure to read it!
>=20
> What I've basically done is switching java keywords being used to c#
> keywords. Now, I know that this is in no way complete _but_ it is a
> start and by having it in cvs others might be tempted to help. I myself
> am a Java developer so this whole new c# thingy is a bit new to me.
>=20
> Patch is attached.
>=20
> So tell me guys do you have any other suggestion as to how approach this
> in a better way or do it the "right way"[TM] ;) As it is now it is more
> or less the generate_code_java.c with a few minor tweaks.
>=20
> What should I do in which order so that we will one day have "complete"
> c# support in dia2code?
>=20
>=20
>=20
> /Richard Torkar

[Dia2code-general] C# support

[Richard T. <ric...@ht...>]

Hi Javier and others!

I have created a generate_code_csharp.c which gives me basic
functionality for generating c# code. (I edited the makefiles and
main.c/dia2code.h/code_generators.h). And I have used it this week for a
project.

It was very easy to do so by using Javier's generate_code_java.c ;)
Very nice code Javier, it is a pleasure to read it!

What I've basically done is switching java keywords being used to c#
keywords. Now, I know that this is in no way complete _but_ it is a
start and by having it in cvs others might be tempted to help. I myself
am a Java developer so this whole new c# thingy is a bit new to me.

Patch is attached.

So tell me guys do you have any other suggestion as to how approach this
in a better way or do it the "right way"[TM] ;) As it is now it is more
or less the generate_code_java.c with a few minor tweaks.

What should I do in which order so that we will one day have "complete"
c# support in dia2code?



/Richard Torkar
-- 
Ph.D. Student
Dept. of Informatics and Mathematics
HTU

[Dia2code-general] functions and procedures

[linutux l. <lin...@ho...>]

Hello.

When I define a operation, how can I make for to convert to Ada95, the 
operations are procedures and not functions?.

thanks.

Sorry for my english.



_________________________________________________________________
Únase al mayor servicio mundial de correo electrónico: 
http://www.hotmail.com/es

Re: [Dia2code-general] Hello

[Robert K. <rik...@ho...>]

Javier,

Welcome back and thank you for the reply. I can totally understand where you 
are coming from.

There are some very nice things about Dia2Code that I would like to point 
out.

1. It's lightweight.
2. The independence between reading the XML file and processing the output 
means that the code generators could work with XMI formats if someone writes 
a different front-end. That was a very nice architectural move on your part.
3. It's not in C++. Some would call that a liability, and it is, to some 
degree, but it also means that the barrier to entry is low (in other words, 
it's easy to contribute.)

The problem with other code generators is that they are dependent upon the 
tool themselves. This can be a good thing for code generators like Rational 
Rose's, which does a superb job of detailing the code, but it also means 
that different code generators produce different output. The nice thing 
about dia2code is that, because of your initial architecture, another parser 
can be thrown in to the front, or additional code generators can be thrown 
into the back, and before too long, voila! A free-as-in-speech code 
generator that will allow you to generate code from various projects and 
different tools into whichever format you like.

Maybe dia2code itself is not right for that kind of vision. Maybe while 
writing this I'm convincing myself of that. But, *that* is my vision. A 
great code generator that can read from any tool I choose and produces the 
code in the format I choose. Integration with other UML projects is good, I 
suppose, but with repeat-forward engineering, you don't necessarily need 
code generation embedded in the tool, just part of your build (make?) 
process.

Anyway, good luck to you, and let me know if you decide to embark on another 
UML code generation project. I would be most interested in it.

Rob

>Hi, Robert, Lothar and all the list subscribers.
>
>I was offline for a while and had just approved Rob's post, which was held
>for approval because of it's size (>40k).
>
>Anyway, it's been a long time (even for dia2ccode standards ;-)), since 
>I've
>done anything (but set up the CVS tree in SF). The last release was done 
>more
>than 7 months ago, so it's time I should speak out my mind.
>
>1. Status of Dia2Code
>
>Dia2Code is a useful little app that can build the skeleton of the classes
>you draw with Dia's UML shapes in a language of choice.  The design was 
>laid
>out without too many technical considerations.  Some things might have been
>good design choices (intermediate representation of the diagram) and 
>another
>ones might have been bad ones (a C function to lay out the code).
>
>Dia2Code has outgrown the little niche it was built to fill up.  When I 
>first
>coded it (last quarter, 2000), there was little choice of CASE tools for
>Linux.  There was ArgoUML but it was, mostly, centered in Java.  Dia was my
>program of choice for the documentation of a OO project I did about that 
>time
>  (which used C++) and d2c mostly was created because of my desire to build
>the skeleton of the code from the class diagram.   Today, we have not only
>good (but propietary) CASE tools that run on Linux (Rose and Together) but
>also a very promising (and GPL-ed) UML diagram editor (called UML).
>
>The experience I've had with Dia2Code  has showed me the big limitations 
>the
>program has.  Basically, it's hard to extend (you need to code a C function
>to generate code for a new language) and it lacks reverse-engineering.  
>Rob's
>latest patch does a nice work to address the latter.
>
>There are some additions/patches I haven't applied to the code, most 
>notably
>Lothar's Perl code generation.  Not because they're technically lacking, 
>but
>because of my laziness/short memory.  Anoher type of patches just tried to
>address the indentaion, which is something I've intentionally left out,
>because there are good code indentators out there (I prefer Artistic 
>Style).
>
>I'd love to see a GPL-ed CASE tools that does welll with round-trip
>engineering and is simple to customize, yet I think Dia2Code is neither the
>solution nor the basis for it.  That said I should add that my willingness 
>to
>maintain it tends toward zero as time passes.
>
>
>2.  What is needed
>
>The experience I've earned  from Dia2Code development and comments from the
>users, and from using (shortly) a propietary CASE tool (Together), shows me
>that there are some key points to address:
>
>a) Flexibility:  Users demanded flexibility in the indentation style., 
>which
>cannot be addressed with the current structure of Dia2Code.  Other 
>parameters
>should be useful (as many of you pointed out), so we should need a few new
>command-line arguments.   A configuration file should be a nice addition.
>
>b) Round-trip:  A good CASE tool should be able to re-create the UML class
>diagram from the modified sources.  Rob's mehthod is a nice start, yet I
>think a template-driven code generation should address both a) and b).  
>Such
>template structure is rather complex to describe here, so I'll omit it for
>the sake of clarity, because it could be a complex issue to tackle.
>
>c) Compatibility:  Not everyone uses Dia to make its class diagrams.  Why
>should somebody who uses Rose/Together care about it?   Answer: XMI 
>support.
>A good CASE tool should support import and export from/to XMI, which is, to
>keep it simple, a standard, XML notation to represent UML class diagrams.
>Chris McGee (Dia2Code developer) has done some work with XMI for the UML
>project, so maybe he can help the most.
>
>d) Dia integration:  A Dia plugin with a nice GUI for all the features is a
>must.  It doesn't need to be a whole program, maybe just a  stand-alone
>program and a Dia plugin that invokes it.
>
>e) Freedom:  Dia2Code was born from the need of a free-as-in-speech tool 
>that
>could do C++ (and Java, and many other languages).  A free program should 
>be
>avaliable from a free to use repository.   SourceForge was beautiful in 
>2000,
>really.  VA Linux was very very promising.  Now, VA Software sells
>propiertary extensions of it and recently, it announced the switch to
>propietary IBM platforms.  Not that they're charging for the service, and
>not that I dislike them.  I'm very *very* grateful to all the VA Linux 
>people
>for their cool service, but we should switch now, as the future of their
>services is not clear.  In 2000 we didn't have a choice of free-as-in-beer
>public repositories, but now we have Savannah, a SF-like site hosted by the
>FSF.
>
>
>3.  Future of Dia2Code
>
>As I stated above, I'm not willing to maintain the project, this is mainly
>because I feel that is now obsolete and because I'm not constant in my
>efforts.  I have a full-time job now (well, it's been a while now) and 
>little
>time left for C hacking.  Candidates (please, just from the current 
>developer
>popultaion) are welcomed.
>
>Patches should go to the patch system of SF, not mailed to me.  Any of the
>current SF-registered developers can apply them to the CVS tree.  I'd try 
>and
>find some time to apply pending patches.
>
>If people still want to continue d2c development, it's OK with me.  I'd
>rather start a new project that has the properties described above.
>
>Hope this was good food for thoughts,  thanks fot the help and patience.
>
>Regards,
>
>   Xav


_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com

[Dia2code-general] Hello

[Javier O'H. <j....@ti...>]

Hi, Robert, Lothar and all the list subscribers.

I was offline for a while and had just approved Rob's post, which was held 
for approval because of it's size (>40k). 

Anyway, it's been a long time (even for dia2ccode standards ;-)), since I've 
done anything (but set up the CVS tree in SF). The last release was done more 
than 7 months ago, so it's time I should speak out my mind.

1. Status of Dia2Code

Dia2Code is a useful little app that can build the skeleton of the classes 
you draw with Dia's UML shapes in a language of choice.  The design was laid 
out without too many technical considerations.  Some things might have been 
good design choices (intermediate representation of the diagram) and another 
ones might have been bad ones (a C function to lay out the code).  

Dia2Code has outgrown the little niche it was built to fill up.  When I first 
coded it (last quarter, 2000), there was little choice of CASE tools for 
Linux.  There was ArgoUML but it was, mostly, centered in Java.  Dia was my 
program of choice for the documentation of a OO project I did about that time 
 (which used C++) and d2c mostly was created because of my desire to build 
the skeleton of the code from the class diagram.   Today, we have not only 
good (but propietary) CASE tools that run on Linux (Rose and Together) but 
also a very promising (and GPL-ed) UML diagram editor (called UML). 

The experience I've had with Dia2Code  has showed me the big limitations the 
program has.  Basically, it's hard to extend (you need to code a C function 
to generate code for a new language) and it lacks reverse-engineering.  Rob's 
latest patch does a nice work to address the latter.

There are some additions/patches I haven't applied to the code, most notably 
Lothar's Perl code generation.  Not because they're technically lacking, but 
because of my laziness/short memory.  Anoher type of patches just tried to 
address the indentaion, which is something I've intentionally left out, 
because there are good code indentators out there (I prefer Artistic Style).

I'd love to see a GPL-ed CASE tools that does welll with round-trip 
engineering and is simple to customize, yet I think Dia2Code is neither the 
solution nor the basis for it.  That said I should add that my willingness to 
maintain it tends toward zero as time passes.


2.  What is needed

The experience I've earned  from Dia2Code development and comments from the 
users, and from using (shortly) a propietary CASE tool (Together), shows me 
that there are some key points to address:

a) Flexibility:  Users demanded flexibility in the indentation style., which 
cannot be addressed with the current structure of Dia2Code.  Other parameters 
should be useful (as many of you pointed out), so we should need a few new 
command-line arguments.   A configuration file should be a nice addition.

b) Round-trip:  A good CASE tool should be able to re-create the UML class 
diagram from the modified sources.  Rob's mehthod is a nice start, yet I 
think a template-driven code generation should address both a) and b).  Such 
template structure is rather complex to describe here, so I'll omit it for 
the sake of clarity, because it could be a complex issue to tackle.

c) Compatibility:  Not everyone uses Dia to make its class diagrams.  Why 
should somebody who uses Rose/Together care about it?   Answer: XMI support.  
A good CASE tool should support import and export from/to XMI, which is, to 
keep it simple, a standard, XML notation to represent UML class diagrams.  
Chris McGee (Dia2Code developer) has done some work with XMI for the UML 
project, so maybe he can help the most.

d) Dia integration:  A Dia plugin with a nice GUI for all the features is a 
must.  It doesn't need to be a whole program, maybe just a  stand-alone 
program and a Dia plugin that invokes it.

e) Freedom:  Dia2Code was born from the need of a free-as-in-speech tool that 
could do C++ (and Java, and many other languages).  A free program should be 
avaliable from a free to use repository.   SourceForge was beautiful in 2000, 
really.  VA Linux was very very promising.  Now, VA Software sells 
propiertary extensions of it and recently, it announced the switch to 
propietary IBM platforms.  Not that they're charging for the service, and  
not that I dislike them.  I'm very *very* grateful to all the VA Linux people 
for their cool service, but we should switch now, as the future of their 
services is not clear.  In 2000 we didn't have a choice of free-as-in-beer 
public repositories, but now we have Savannah, a SF-like site hosted by the 
FSF.


3.  Future of Dia2Code

As I stated above, I'm not willing to maintain the project, this is mainly 
because I feel that is now obsolete and because I'm not constant in my 
efforts.  I have a full-time job now (well, it's been a while now) and little 
time left for C hacking.  Candidates (please, just from the current developer 
popultaion) are welcomed.

Patches should go to the patch system of SF, not mailed to me.  Any of the 
current SF-registered developers can apply them to the CVS tree.  I'd try and 
find some time to apply pending patches.

If people still want to continue d2c development, it's OK with me.  I'd 
rather start a new project that has the properties described above.

Hope this was good food for thoughts,  thanks fot the help and patience.

Regards,

  Xav

Re: [Dia2code-general] Re: [Dia2code-general] Keeping old code while regenerating (con't)

[Robert K. <rik...@ho...>]

> > (Is anyone else reading this? Hey Lothar, what if you and I are the only
> > ones on this list?)
> >
> No, you're not the only ones :)
>
> I've got a small patch for PHP generation (putting stuff in correct order
+ file headers). I sent it to the author but no reply.
>
> Your ideas are of great interest to me, but I've currently no time to work
on it (I don't even have my computer, thus, no work possible!).

OK, all in good time. I wonder if he'll show up in September when school
starts again!

Rpb

[Dia2code-general] =?iso-8859-1?Q?Re:_[Dia2code-general]_Keeping_old_code_while_regenerating_(con't)?=

[<jea...@la...>]

> How about just emailing it directly to me (instead of 
irritating everyone 
> else.)
> 
> (Is anyone else reading this? Hey Lothar, what if you and I 
are the only 
> ones on this list?)
> 
No, you're not the only ones :)

I've got a small patch for PHP generation (putting stuff in 
correct order + file headers). I sent it to the author but no 
reply.

Your ideas are of great interest to me, but I've currently no 
time to work on it (I don't even have my computer, thus, no 
work possible!).

-- 
JB

Acc=E9dez au courrier =E9lectronique de La Poste : www.laposte.net ; 3615 LAPOSTENET (0,13 =80/mn) ; t=E9l : 08 92 68 13 50 (0,34=80/mn)"

Re: [Dia2code-general] Keeping old code while regenerating (con't)

[Robert K. <rik...@ho...>]

>Am Die, 2002-07-30 um 18.39 schrieb Robert Konigsberg:
> > > >    public void suspend (  )
> > > >    {
> > > >        // ## Implementation preserve start class 
>method.suspend@void@@
> > > >
> > > >        deactivate();
> > > >        active = false;
> > > >        // ## Implementation preserve end class method.suspend@void@@ 
> >
> > >  }
> > Actually it would be easier to implement other languages using this
> > approach. If we needed to perform a diff/patch, we'd need to be > able 
>to parse function declarations, compiler switches, and so > forth, for 
>every language. This way,
>why?
>
> > all
> > you need to do is know what comment-structure is used for the language, 
>and
> > work
> > >from there.
>youre right, thats an easy way.

And that's why! :)

Seriously, you need to parse function declarations in order to identify 
which functions you are reading. For languages where the same function can 
be listed several times with different parameter lists (i.e. C++, Java), you 
need to be able to parse the parameter lists.

Trickiest of all is knowing when the function ends. For languages with 
compiler definitions (C/C++) you need to parse through those. Yuck, yuck, 
yuck. And with Perl you need to be able to parse the language to work 
through HERE documents, and so forth. Comments are pretty-much the easiest 
way to go.

> > Yes, that's what I wanted to know. WHO HAS OTHER CHANGE TO DIA2CODE?
>
>i mentioned the generate_perl.c which i wrote, but there is work left of
>course...

How about just emailing it directly to me (instead of irritating everyone 
else.)

(Is anyone else reading this? Hey Lothar, what if you and I are the only 
ones on this list?)

Rob

_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com

Re: [Dia2code-general] Keeping old code while regenerating (con't)

[Lothar <lot...@gm...>]

Am Die, 2002-07-30 um 18.39 schrieb Robert Konigsberg:
> >if you think of chances to contact him, to add the code to the tree,
> >mmmmh no idea. Post it here?
> 
> Dunno. Apparently the last time he logged into SourceForge was approximately 
> June 11th. Maybe he's a student on summer break. Either way, maybe we should 
> hold off.
> >
> > >    public void suspend (  )
> > >    {
> > >        // ## Implementation preserve start class method.suspend@void@@
> > >
> > >        deactivate();
> > >        active = false;
> > >        // ## Implementation preserve end class method.suspend@void@@ >   
> >  }
> >
> >what about a diff/patch approach? Perhaps other langs could be covered
> >easily this way.
> 
> Actually it would be easier to implement other languages using this 
> approach. If
> we needed to perform a diff/patch, we'd need to be able to parse function
> declarations, compiler switches, and so forth, for every language. This way, 
why?

> all
> you need to do is know what comment-structure is used for the language, and 
> work
> >from there.
youre right, thats an easy way.

 
> >if you mean changes, i didnt get it:)
> >to add code dia dia2code?
> Yes, that's what I wanted to know. WHO HAS OTHER CHANGE TO DIA2CODE?

i mentioned the generate_perl.c which i wrote, but there is work left of
course...

so far 
lothar

[Dia2code-general] Keeping old code while regenerating (con't)

[Robert K. <rik...@ho...>]

>if you think of chances to contact him, to add the code to the tree,
>mmmmh no idea. Post it here?

Dunno. Apparently the last time he logged into SourceForge was approximately 
June 11th. Maybe he's a student on summer break. Either way, maybe we should 
hold off.

>
> >    public void suspend (  )
> >    {
> >        // ## Implementation preserve start class method.suspend@void@@
> >
> >        deactivate();
> >        active = false;
> >        // ## Implementation preserve end class method.suspend@void@@ >   
>  }
>
>what about a diff/patch approach? Perhaps other langs could be covered
>easily this way.

Actually it would be easier to implement other languages using this 
approach. If
we needed to perform a diff/patch, we'd need to be able to parse function
declarations, compiler switches, and so forth, for every language. This way, 
all
you need to do is know what comment-structure is used for the language, and 
work
from there.

>if you mean changes, i didnt get it:)
>to add code dia dia2code?
Yes, that's what I wanted to know. WHO HAS OTHER CHANGE TO DIA2CODE?

Thanks, Rob

_________________________________________________________________
Join the world’s largest e-mail service with MSN Hotmail. 
http://www.hotmail.com

Re: [Dia2code-general] Re: Dia2code-general digest, Vol 1 #14 - 2 msgs

[Lothar <lot...@gm...>]

Am Don, 2002-07-25 um 23.15 schrieb Robert Konigsberg:
> >this is mainly a task for dia, or? AFAIK it is not possible to add code
> >to a method.
> >Im not sure, how far you can customize the properties-dialog for a dia
> >object.
> Lothar,
> 
> As you can see by my prior email, I actually did most of it already.
> 
> I've only been doing this for a week, so I don't know if Javier is active or 
> not, but I do know that I sent an email message to him with no reply, and a 
> message to the list that required the adminstrator to approve (I assume it's 
> him.)
> 
> How many other changes do people have?

if you think of chances to contact him, to add the code to the tree,
mmmmh no idea. Post it here?


>    public void suspend (  )
>    {
>        // ## Implementation preserve start class method.suspend@void@@
>
>        deactivate();
>        active = false;
>        // ## Implementation preserve end class method.suspend@void@@ 
>    }

what about a diff/patch approach? Perhaps other langs could be covered
easily this way.
 
if you mean changes, i didnt get it:)
to add code dia dia2code?

lothar

[Dia2code-general] Re: Dia2code-general digest, Vol 1 #14 - 2 msgs

[Robert K. <rik...@ho...>]

>Hi,
>
>i dont know if this list or project is still alive?
>but, anyway, there is someone who is interested in dia2cod.
>I send a generate_code_perl.c to the author, no reply. Perhaps the code
>was too ugly:)
>
> >     Also, are you looking into code-regeneration -- the ability to 
>maintain modifications to a class? That would be an interesting project, 
>for sure.
>
>this is mainly a task for dia, or? AFAIK it is not possible to add code
>to a method.
>Im not sure, how far you can customize the properties-dialog for a dia
>object.
Lothar,

As you can see by my prior email, I actually did most of it already.

I've only been doing this for a week, so I don't know if Javier is active or 
not, but I do know that I sent an email message to him with no reply, and a 
message to the list that required the adminstrator to approve (I assume it's 
him.)

How many other changes do people have?

Rob

_________________________________________________________________
Chat with friends online, try MSN Messenger: http://messenger.msn.com

[Dia2code-general] Dia2Code - Ability to regenerate framework without removing source code mods

[Robert K. <rik...@ho...>]

Hello all,

SUMMARY
I have made modifications to dia2code.c, dia2code.h and =
generate_code_java.c so that you can generate Java code, add code to the =
stubs, modify the model, regenerate the code and you don't lose any of =
your stubs.

Of course, I used a commenting mechanism similar to other code =
generators, which means that it's not the nicest looking thing, but it =
works, and if you have an editor that colors based upon syntax, it's =
pretty easy to use.

It means that any code which you want to maintain must remain within =
these "preserve comment blocks", i.e.

For example:

    public void suspend (  )
    {
        // ## Implementation preserve start class method.suspend@void@@=20
        deactivate();
        active =3D false;
        // ## Implementation preserve end class method.suspend@void@@=20
    }

Of course, I'd recommend a series of appropriate command-line arguments =
to enable and disable parts of this functionality.

Additionally, if you remove a method or change it, it won't delete the =
code immediately; it'll warn you and move it into a comment block.

STATE OF CODE
It's functional. It can be implemented in the other languages; it's =
pretty straightforward. It could use a couple of mods. Specifically:
1. I'm not sure how easily this will work with the .cpp and .h =
situation. This might be a couple hours work to get this to work.
2. It's "//"-centric. This needs to be addressed.
3. It really should have accompanying command-line parameters.
4. The interface is a little kludgy, I can clean that up. But, I can get =
this code to someone to look at merging into the main branch.

ALL THAT SAID,
What should I do?

Rob

Re: [Dia2code-general] Re: Auto-indentation / std interface

[Lothar <lot...@gm...>]

Hi,

i dont know if this list or project is still alive?
but, anyway, there is someone who is interested in dia2cod.
I send a generate_code_perl.c to the author, no reply. Perhaps the code
was too ugly:)
 
>     Also, are you looking into code-regeneration -- the ability to maintain modifications to a class? That would be an interesting project, for sure.

this is mainly a task for dia, or? AFAIK it is not possible to add code
to a method.
Im not sure, how far you can customize the properties-dialog for a dia
object.


lothar

 

[Dia2code-general] 2 Errors found in C++ code generation

[<dia...@bi...>]

Hello dudes...

I found 2 misbehaviours in C++ code generation. 

When creating a "uses" relationship between two objects, the include of the 
other objects header file is made on the wrong side (it should be on the 
side without arrowhead) 

When using aggregation (White ("empty") diamond), an object is created in 
the objects attributes, when using composition (Black "filled" diamond) a 
pointer to an object is created in the object attributes. AFAIK this should 
be the other way round (that's the way I understood the 
UML-specifications... compsition is stronger than aggregation) 

gtx
michael