Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - Martin Merkel

Pages: 1 2 3 [4]
46
Automation Interface, Add-Ins and Tools / Re: Importing C source code
« on: August 16, 2010, 08:01:25 am »
Quote
Thanks for the hint, that does the job.

Martin

Oops, at problem still. Obviously EA does not give me control of the sequence in which structure elements appear in the structure. I've taken the header above and imported it into EA, than added one variable at the end in one of the structures. When forward synchronizing EA however always puts it to the front of the structure, regardless to where in the structure I move the entry in the attributes dialog.
I also did another test with defining a new struct in the model, assigning a source file and the generating the code. After the initial generation, the sequence of my structure entries reflect the sequence of the attributes in the model element, if I lateron modify the sequence of the attributes in the model this is not propagated to the source code on the next synchronization.
Any hints?

47
Automation Interface, Add-Ins and Tools / Re: Importing C source code
« on: August 14, 2010, 01:01:09 am »
Thanks for the hint, that does the job.

Martin

48
Automation Interface, Add-Ins and Tools / Importing C source code
« on: August 12, 2010, 09:41:41 pm »
I have a couple of C structures that features conditional compilation, and that I would like to reverse engineer with EA. The problem is when I later generate source code from the diagram that was constructed while importing the code, I loose the conditional compilation part of it.

Example:


typedef union
{
    uint32_t    ui32;
    uint16_t    ui16;
    uint8_t     ui8;
}t_DATA;

typedef struct
{
    uint32_t    u32_id;
    t_DATA      u_data;
#ifdef (FLAG)
    uint8_t     u8_flag;
#endif
}message;


Is there any way to make Enterprise Architect understand such constructs?

49
Oups,
Looks like I oversimplified the code a bit too much such that instead of the existing code I had exiting code....

Anyway, here's snippet for which EA (8.0b3 since today) produced an error on importing:

01:  #define CHAR            char
02:  #define ROMCONST    const
03:
04:  typedef struct info_a
05:  {
06:    const char *a_name;
07:  } INFO_A;
08:
09:  typedef struct info_b
10:  {
11:    ROMCONST CHAR *b_name;
12:  } INFO_B;

Message is:

There was an error parsing Y:\test.h on line 12.  Unexpected symbol: }

In fact it wasn't the '*' that caused the error message, but rather the ROMCONST qualifier define (CHAR alone is accepted), which we use to abstract compiler specifics. We need to build our code under a dozen or so different compilers. Given this in real code the define may look different from target to target.

Identical results I get when using a #define VOLATILE volatile as qualifier before the char. Where could I teach EA to interpret such qualifiers, or at least to ignore them.

Thanks,  Martin

50
Im trying to model code that will be implemented in good old plain C. For a start I did some tests trying to import exiting (but heavily simplified) code via Code Engineering->Import Source Directory. I run into problems however with code such as 'char *pc_name;'  I get error messages such as 'Unexpected symbol: *'

I tried to browse the forum and the white papers section on topics related to reverse and forward engineering of code, however all articles an posting focus on object orientated languages and not on how to proceed with the 'classical' languages such a C.

Is there any additional background information on possible limitations, or has somebody gather experience on possible limitation in this respect.

I'm currently using EA 8.0 beta 2

Thanks for any input

51
Hi, is there any news on the MDG Integration for Visual Studio v4.0 which was released as beta 1 back in January this year?

52
Basically I try to implement a profile that allows my modeling safety relevant software developments according to IEC61508. On of the things that one is supposed to do is a formal documentation of the requirements tracking and design process, including who performed the review and who finally verified the review. For each of these process only specific People will be 'qualified'. These have to be documented including their affiliation etc. I thus would have like to include the corresponding information into my people definition, and then later on in for example my FunctionRequirement stereotype add attributes according to my tagged values.
This could look something like below:

Type=Enum;
List=Authors;
Role=Project Manager;

As you mentioned this is currently not possible, so I will just use the Authors keyword for the time being. It may however be an interesting feature for further revisions of Enterprise Architect to support such functionality. Also the possibility to enter company data corresponding to individual People may be a useful feature.

53
Thanks for the reply, though it does not appear to work as role types do not seem to be UML sterotypes.
Generally, when editing people working on a project, I'm missing that I can select from a predefined set of roles for People, and not for Resources (where on the other hand I can add company affiliation etc. to a particular resource. I have problems understanding the differences here between People and Resources.

54
Hi, maybe somebody can help:
Is there any way to filter according to specific role types while defining tagged values? I know of

Type=Enum;
List=Authors;

but I would rather prefer to allow selection from a list that corresponds  to a specific role type, such a Project Manager or Developer.

55
Similar context, question from a beginner: When I export a requirements profile created based on the description the the Sparx Systems white paper, type, values, default, etc are not written into the generated UML file. What do I do wrong here?

Thanks, Martin

Pages: 1 2 3 [4]