|
| Re: numeration of requirement id [message #1827871 is a reply to message #1827865] |
Tue, 26 May 2020 07:09   |
|
Hi John,
in principle YES it is possible. But ...
A design principle largely used with Papyrus are pre- und post-action hooks. A piece of Java code linked as post-action might alter the values of an element created.
Even if no hook exists in that very case, Eclipse is FOSS ;-)
But is it worth the effort.? So do a business case first.
I my self so far haven't used that feature.
/Carsten
|
|
|
|
| Re: numeration of requirement id [message #1827877 is a reply to message #1827875] |
Tue, 26 May 2020 08:13 |
|
First off all, I try to get rid of SysML whenever possible. One reason is in SysML a sysml::Requirement is a specialization of uml::BehavioredClassifier. A requirement is neither classifying anything nor has an inherit behavior. That is by far not the only issue I have with SysML.
Back to the topic:
I personally prefer stories (or scenarios) over atomic IREB requirements. With atomic requirements one always faces the problem that the requirements engineer dissects her/his thoughts / intents into several thousends of atomic requirements. The systems designer then has to reverse engineer these thoughts / intents based on these huge pile of requirements. In the last 20+ years I faced many times the situation that the design team perfectly covered all requirements but designed something completely different than the REs intended. In a recent case they designed an algorithm to figure out and compensate temperature sensor failures. What was intended was that the inverter switched between breaking via low side and high side short circuit and vice versa depending on which side is significantly colder. Well, all requirements have been perfectly covered. The REs confirmed this. But something completely unintended has been designed.
A story is an annotated uml::UseCase contrained by uml::Constraints and backed by an uml::Activity.
OK, if I have to use requirements I personally keep them behind DOORS ;-)
/Carsten
|
|
|
| Re: numeration of requirement id [message #1827885 is a reply to message #1827875] |
Tue, 26 May 2020 09:35 |
|
> How do you cope with inserting reqiurements between existing requirements afterwards.
> Do you change all following IDs by hand ?
I consider the ID as a mere unique identifier. As long as the ID's values are unique all is fine.
Given that, I do not renumber any ID at all.
Why an ID number bears no semantic at all. It is a mere sequence of digits.
/Carsten
|
|
|
| Re: numeration of requirement id [message #1828167 is a reply to message #1827885] |
Wed, 03 June 2020 08:17 |
|
Hello,
you can also use the requirement table (or other table that you configured to display requirement id column) to set the id of your requirement. The table will be filled automatically, then select define the id of the first requirement of the table, then select the cell, a cell selection looking like Excel will appears, and you can extends this cell selection to fill others cells increasing the id value.
You will find the description of this feature in the Eclipse Help Contents : Papyrus Guide -> Using Tables -> General -> Table Features -> Fill Cells
/Vincent
|
|
|
| Re: numeration of requirement id [message #1828170 is a reply to message #1828167] |
Wed, 03 June 2020 08:41 |
|
Hi,
with both automatic or manual numbering you face the problem of breaking references. In many engineering projects UIDs are used to reference information items of any kind.
Even if it is just in an eMail, like: "Joe, could you please have a look at Rq-Diag-003653." and the 003653 gets altered you face a problem.
/Carsten
|
|
|
Powered by
FUDForum. Page generated in 0.03010 seconds
Search
Help
Register
Login
Home
