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 - Shegit Brahm

Pages: [1] 2 3 ... 5
1
Uml Process / Re: Signals - "Details" tab not displayed
« on: December 05, 2018, 12:29:52 am »
The details page exists to provide other options that aren't available for other types. It's only an artifact of history that it provides access to the Attributes and Operations. It's never been the primary way to access those windows.

Well, sounds like the UX should be evaluated and the UI redesigned.

I like to have a consistent one and more than one way to achieve something via GUI.  Because I think counterclockwise where others think clockwise. And because "this is the obvious way" is a very subjective view based on what I experienced in the past.

2
Suggestions and Requests / Re: [Forum] how notify spam
« on: November 30, 2018, 07:00:04 pm »
Maybe if I ban my own IP address, the problem will go away. For me, at least  :)
:D ;D 8)

PS: thanks for the other hints, will follow them

3
Suggestions and Requests / Re: [Forum] how notify spam
« on: November 29, 2018, 06:49:10 pm »
@Moderators: would you encourage us users to hit the "notify" button (like Geert did) or would you agree qwerty's assumption of "they are cleaned next monday"?

Thanks any idea which user's help would it make easier for you to get rid of it, Shegit

4
Hi Martin,

cross reading your link points me back (& supports me) to my initial question:

Do you want to model the process of decisions made (might call it the decision finding system) or do you want to model the result (might call it the bank system)?

In both cases you need UseCases and requirements to describe the system.

Where is the difference? For what the system is used.

In first case:
You describe a requirements management tool (RMT). This tool is able to track down how a decision made regarding the "bank system" (BS).

In second case:
You describe the bank system (BS). This tool will be used by John Does, e.g. to do account transfers.

That means, that the requirements for the RMT reflect the business process, how decisions are made to create the BS.
While the requirements for the BS are these ones you describe as
This decision will impact existing requirements (use cases, functional, etc.),  processes(some will be changed, other removed and new one created), systems, infrastructure and more. We want to be able to trace back the changes to the decision that is at the root. It could also be use to give feedback to management of the impact of the decision before proceeding.


So as far as I understood/ assume: you have two systems: one is for tracking requirements that are the base for a system.

That means:

The usecase-requirements-description for the RMT contains the made decisions only as "artifacts to be stored properly". And it results in code, that is the RMT.

While the usecase-requirements-description for the BS contains e.g. account details as "artifacts to be stored properly". And it results in code, that is the BS.

And with a good requirements management tool you can do impact analysis, you can track down a system requirement to a stakeholder wish and also to a toplevel goal of the entire BS.
And you can use this RMT again to design the BS2 - independent from the further development of BS, just the process how to develop both BS is the same.

Coming back to your initial question: well, you want to model a RMT, which artifacts & database objects are used to create a BS - and to tell the developer of the BS how much impact a change in stakeholder requirements has.

Just don't mix requirements (for RMT) with requirements (for BS).

So a requirements manager would use the RMT to tell the bosses how the BS looks like.

Cheers, Shegit

Cheers, Shegit

5
Hi carles,

as far as I experienced EA, there is an important difference between a "classical programming merge" and the EA merge.

in my observation: EA says, a merge is an import. There is no "take their side / take my side" decision.

Due to the fact: you have a model here, not code lines. See answer by Thomas Kilian https://stackoverflow.com/questions/32362684/work-arounds-for-the-lack-of-version-control-branching-support-in-enterprise-arc/32370908#32370908

Despite your detailed description what you did, I was to lazy to repeat it. Sorry.
Just I expected to get Model A setting into B after your EA merge "A into B".

_the_ most important thing in EA is the GUID.

So in any case I make an update of an element, the EA will look for the GUID and "rebuild" found element.

That works with XMI-Import (where I can easily manipulate any information),
it works with reusable assets (RAS) and also with baselines:

That way I got entire packages "restructured" and elements lost, due to some tiny GUID on an existing element.

HTH, Shegit

PS: There is this feature "baseline" which allows you to take over only parts - of course, a connector needs both elements and the visibility of an element this one. Most of all: I found only a GUI-approach to do a baseline comparison.

PPS: LieberLieber sells with Lemontree a comparison tool that offers a nicer GUI approach and promises some automatic merge. Hence they require that you have the entire .eap file as one file under SVN, not each package inside EA (obvious effects: you up/download allways everything & you can get in conflict with other users)

(Yes, I "exvacte" it, because my findings are new for me, I didn't find that much answers on my own searching, so I "force" you to see my "share")

6
As I have only UML background, I disagree the idea that requirements would be derived from decisions.

In "my world" a decision is like what KP said: e.g. splitting up control flows based on conditions.

And a requirement tells what a system should / must do. Not how. This is done by e.g. Activity Diagrams.

So regarding your "business decisions" that is for me you document a process of what should be done.
Which results in requirements - the thing is: do model the process or the result?

Because for the process, I would have an (UML) activity that contains some actions.
One action element would have a control flow to the decision element which is named after your "business decision" (like: what would we do now?).
Follow up would be e.g. 2 actions: one says "keep it as it is", the other would say "change x" associated with an artifact "new requirement".

Just this "new requirement" would not be a requirement of the system the activity describes. For me it would be a requirement for a different system.

So I would have to find something else than the UML requirement & decision element for your case, I guess.

[Update:] Could you share a screenshot, please?

HTH, Shegit

7
Sounds like it would do harm if you actually used it.[...]

Right now no error occured.
After this error report started, I set some tagged values, registered them, imported these assets somewhere else - and these tagged values got imported as well (and stored in the origin database).

Maybe I could create some own types and do more complicated stuff with tv to see if it gets broken.
On the other hand I do almost nothing in automation, so my explorative tests are clearly limited.

Cheers, Shegit

8
What build of EA are you using?

I'm using build 1351. (Version 13.5.1351)

Sounds I can ignore it as it does no harm, just thinning my nerves as long as I think it harms?

9
Hi there,

while exploring the world of RAS (Reusable Asset Service), I stumbled upon this error while I do the mandatory dependency check before registering an asset:

Quote
Microsoft OLE DB Provider for ODBC Drivers [-2147217865]

[MySQL][ODBC 5.3(w) Driver][mysql-5.7.24-0ubuntu0.16.04.1]Table 'EA_RAS_Client.t_taggedValue' doesn't exist

the creation code says:
Code: [Select]
CREATE TABLE `t_taggedvalue`
(
`PropertyID` VARCHAR(40) NOT NULL,
`ElementID` VARCHAR(40) NOT NULL,
`BaseClass` VARCHAR(100) NOT NULL,
`TagValue` MEDIUMTEXT ,
`Notes` MEDIUMTEXT ,
CONSTRAINT `pk_taggedvalue` PRIMARY KEY (`PropertyID` ASC)
)
ENGINE=InnoDB
DEFAULT CHARSET utf8
COLLATE utf8_general_ci

;

This page https://dev.mysql.com/doc/refman/5.7/en/identifier-case-sensitivity.html indicates, that there might be a problem with different OS and their settings - just the EA script uses only lower case.

And the DB also shows me the table, all lower case.

Any idea what I could do about it?
Searching for first error line brings up (mostly) "you are connecting to a mispelled table" - so the mysql statement might be all.
Just ... the table created by Sparx' script is lower case.

I would claim that after the initial setup, there was no such error.
Maybe it was only an error on one of three DB's and while I moved assets from a to b to c and vice versa - this error got "exported"?

Or is it a bug inside RAS feature?

Thanks, Shegit

10
General Board / Re: How to setup a RAS (Reusable Asset Service)? [solved]
« on: November 15, 2018, 11:40:33 pm »
    @Graham_Moir: thanks for your encouragement.

    I digged deeper and will post some major "ah"-moments.

    assets:
    • smallest unit: empty package
    • => package is general unit
    • => no single element can become an asset
    • latest version is the highest version number EA says - not always the newest
    • change an asset?  - make a new version!
    • merge: only from baseline (e.g. one asset version) into local model - there is partial / elementwise merge=import possible (local delete as well)
    • access different version: the dropdown does appear only if you know to look in the version column
    • same guid? - "crashes/ mixes up" more or less like an xmi import with same guid on different elements

    dependencies
    • will be treated as separat assets - and dependent version will be saved, too

    storage:
    • password(s) can be set - once all passwords are gone / never set => no new password(s) can be set
    • as long as you have access to a storage, you can easily copy it
  • lost password? guess this is the encrypted version (xmi export):
Code: [Select]
<code header1="dY1h8mpERIOE5cRGig3Leiq+RWBnsPe1BbwouUEf6TE=" header2="dY1h8mpERIOE5cRGig3Leiq+RWBnsPe1BbwouUEf6TE="/>

    be careful
    • you update two assets and the second is dependency of the first and you register new version of first asset?
      • please avoid using "use latest version in registry"
      • that would ignore your current local changes of second asset
      • untick it and register second asset with a new version from local project
      [li] take a look, if the "latest version" is really the newest - the dropdown has them in created-order

    BUG
    • the version name permits these characters <>;:“/\|‘,?*
    • well, it checks in advance to registering on these only
    • it DOES NOT check on any non-english characters BEFORE saving into registry server
    • it does realize this error few seconds later while the storage got reloaded to show changes
    • that crashes entire registry server regarding read and write (all storages)
      • I was lucky and could just delete the storage in registry server ea model (because nothing else was in it)

    HTH, Shegit

    11
    General Board / Re: How to setup a RAS (Reusable Asset Service)?
    « on: October 11, 2018, 01:13:48 am »
    Was easier than expected.

    the "registry server" is nothing more than a simple EA cloud connection with a model stored as real DBMS instead eap.

    That means:
    • set up a cloud based EA model.
    • have your second model at hand (with sufficient rights)
    • go to Publish -> (model Exchange) -> Reusable Assets-> Registry: select the EA model set up in first step
    • now you are connected to the central registry server


    That said - I can't tell you right now the sufficient rights to do what inside RAS etc.

    Assets can become what I create locally and "register"=export them into the cloud model.
    And importing from cloud model creates local copies.

    As far as I digged until now: GUIDs will be "central" at the cloud model and whereever you import from there, EA will (try to) set the central GUID.
    Which implies: can't tell you now what happens if I have a local GUID for a different element and the import it from cloud.

    HTH,
    Shegit

    12
    General Board / How to setup a RAS (Reusable Asset Service)? [solved]
    « on: September 12, 2018, 09:26:22 pm »
    Hi,

    this RAS feature sounds nice. So my object is to find out - what is possible, which (technical) limits exist etc.
    (Have EA Enterprise 13.5 at hand)

    I started reading on Sparx' page to get started (setting up) http://www.sparxsystems.com/enterprise_architect_user_guide/13.5/model_repository/setup_asset_service.html

    My problem starts in the second sentence of the first step (identify registry):
    - The Registry server holds the Storages of the Reusable Asset Service.
    - It will have been previously configured by your system administrator, to be accessed through a Cloud Connection.

    Because - I am in that case the system administrator and - I struggle to find any "next" page what I need to do to get a "registry" up and running.

    I have a server, where the Sparx EA service is running.
    I can create a database (MySQL), initialize it as an EA repository.
    Take the connection details, open an local EA on a client machine and connect. And model my way through.


    So my first question is:
    - Which components must be up and running to create a registry server? (And how do I connect to it?)

    I am able to open the RAS view on my local EA, right now it is empty.

    Than there is the second part:
    http://www.sparxsystems.com/enterprise_architect_user_guide/13.5/model_repository/set_up_registry_password.html also notes that I should do it on the Registry Server
    - How do I get there?

    After anything is up and running, I guess this is more easy:
    - Do I create a storage from my local machine? The part of "Create Storages" implies it.

    Many thanks,
    Shegit

    13
    Hi Patrick,

    thanks for your offer.
    In between I downloaded Yakindu EA-Bridge and checked the tutorials, worked in over 90 % of the steps.

    I was not able to connect to a selfmade eap file and my focus is in the first step not a full semantic check. I thought I start low to just read EA elements, make them in Eclipse diagrams visibile and manipulate them. And vice versa.

    I got the example from Sparx' automation page running and load a small eap file as Repository class into eclipse - however that is far from "import an UML model".

    If I understand you correct, you say that your current work will result in something like this?
    In that case I'm looking forward to your personal message in this forum to talk about how I might get hands on your new feature.

    Thank you very much,
    Shegit

    14
    Hi there,

    I have an EA 13.x with MDG to evaluate how to combine an EA model with Eclipse (photon), like the UML models or code stuff.

    My aim is to get a vice versa connection between these two "UMLable" platforms. To add stuff here and edit it there and an update has both on same level.

    My (main) question is:
    • What did / do you use successfully and which way?
    • What would you recommend to use / have a look at?

    The build-in extension from Sparx cannot cope with declarations like "static class Swapper implements Sorter.Swapper" - EA kills "Sorter." which is not nice.  And it shows eclipse warnings as errors, at least it builds the project anyway. And if I change code, the build transfers it - it is just not offered while using "merge with eclipse".


    http://www.sparxsystems.com/forums/smf/index.php/topic,37328.msg234218.html#msg234218 pointed out how to get the feature inside EA running to get a connection itself to eclipse (copy sparx jars into eclipse folders). It just did not appear inside eclipse gui - only under "installed software - plugins" are two mentioned.
    The official guide http://sparxsystems.com/products/mdg/link/eclipse/ does not tell anything that or how I could use it from eclipse side. And it is from 2008.

    Regarding this post http://www.sparxsystems.com/forums/smf/index.php/topic,39647.msg244365.html I tried also Luna (=4.4), but behavior was the same with photon. (mentioned above)

    openArchitectureWare with EclipseLink like suggested here http://www.sparxsystems.com/forums/smf/index.php/topic,28169.msg212539.html is no option any longer since it is discontinued in 2008 with 4.3.1 and no download available - and I don't understand how the "successor" Xtext and Xpand fill the "gap" between EA and eclipse. They seem to focus on DSL building instead.

    I tried to get in contact with the guys from modelbus (see http://www.sparxsystems.com/forums/smf/index.php/topic,30833.msg223984.html), but they lack time to answer. Just their descriptions sound more than wanted.
    The Yakindu EA Bridge I still need to look at. As well as on "export UML to eclipse" suggestion http://blog.lieberlieber.com/2012/11/13/export-your-ea-models-for-eclipse/.

    Thanks for reading and some ideas,
    Shegit Brahm

    15
    Hi there,

    I try to generate a documentation on package_1 by using virtual documents, see project browser structure:
    • Package_1
      • UC diagram
    • (...)
    • Package_5
      • sub package_5.1
        • sub sub package_x.y.z
          • UC1
            • AD (Composite of UseCase)
              • Activity
                • Action1
                • (...)   
                • Final
                • Initial
          • UC2

    Relations
    • UC diagram - UC: UC imported as “Link” in diagram
    • UC - AD: AD is composite child diagram of UC, stored in same package as UC
    • AD - AD elements: AD elements have the activity as parent and are stored in same package

    There's need to get UC diagram, UCs, AD, activity, action, (…), initial and final included in that documentation.

    The first approach to get that done was to use template and fragments. The resulting documentation contains UC diagram, UCs, but neither AD, activity nor anything being a child-child of the UseCase.

    So I turned to virtual documents. In order to grab the missing AD and its diagram elements and get them reported, I tried to create a model search applied as tagged values of a model document.

    But it turned out that my created model search doesn't work. I think, my approach of implementing a model search into a model document isn't correct.
    Beside any general solution to my context I think I need to understand first:
    How to use model search in model document?
    Where do I have to write what? The modal search as a query is a custom sql of a fragment within the template that is chosen for the modal document?

    I was unable to use the help in http://sparxsystems.com/forums/smf/index.php?topic=6152.0 to get a working virtual document that uses a modal search. The semi-automated approach published there by Geert is not “copyable” for me because I need to start on a package, not a diagram.

    Thanks,
    Shegit


    PS: If you prefer to revitalize the old thread as it seems to be the topic: short notice and I will to so.

    Pages: [1] 2 3 ... 5