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 - andykennelly

Pages: 1 [2] 3 4
General Board / Automatically creating diagrams
« on: August 23, 2013, 12:20:42 am »
I know I can easily import a list of elements using the CSV import function.

And I know I can even create links as part of the import if I use scripting. That would then position me to create diagrams with linked elements when subsequently I drag/drop the elements onto a diagram.

I'm now wondering if it's possible to automatically create those diagrams, e.g. for each folder created during the import a diagram is created with the relevant elements and their links - if this is technically possible, I'm seeing a very complex piece of scripting to make it happen!

Maybe the nearest I can get to this is to manually drag/drop and then use Diagram | Layout Diagram.

Any views? Is this somewhere anybody else has been?

General Board / Impact analysis - objects rather than links
« on: July 16, 2013, 01:05:05 am »
I can drop an element as a link on a digaram, but only once. If I want that same element to appear more than once on the same diagram I have to drop it on as an object.
While this might be correct UML modelling, my questions about this are:-
1. If a particluar element is used a number of times in a diagram (or across several diagrams in the same folder/project) the project browser gets quite cluttered. Is there a way to hide these (assuming that's not bad practice?!)?
2. I can't find a way to trace back to the actual element, e.g. alt+G whilst cursor is on the digram object results in the project browser object being highlighted - it doesn't position on the element of which the object is an instance.
3. Similarly, View/Traceability doesn't "trace back" to the element itself.
I can't see how, in a scenario where I've had to use objects, that I can do full impact analysis on an element (i.e. if an element is stored in Folder 1 and modelled as an object in Folder2, how do I find out where that element is used in other folders?).

I'm sure I'm missing something basic.

General Board / Mapping SOA service(WSDL)  fields to a class model
« on: June 24, 2013, 07:25:24 pm »
In my organisation we’ve created component and interface elements to  model our SOA service estate. As part of that, we’ve even imported WSDL’s into the relevant part of the model.

Now that’s done, we are modelling a system that will be technically implemented over a MVC pattern on a .net platform. We’ve created class diagrams to represent the application model onto which we now want to map to the various SOA services required to populate that model (i.e. Customer class will be linked to SOA service getParty, getPartyAddress etc.).

So far, so good …… but we’re now wondering if it’s possible to go one step further and map the fields within the SOA services to the relevant attributes of the class, i.e. CustomerPrimaryName attribute of Customer class is populated from getParty.primaryName. We’ve haven't been able to work out if this is possible using Sparx, but since the tool enables us to import the WSDL and since it seems such a useful/obvious thing to want to do, we can’t help but feel somebody must have been down this road before??!!! Maybe using a plug-in?  Maybe it’s possible for Sparx to generate some .net code for the transformation?

(My example here is simple, of course …. things might be trickier where we have to combine several service fields to populate a single class attribute).

Has anbybody been here before?

General Board / Re: 'Require user lock to edit' and group locking
« on: June 19, 2013, 11:37:10 pm »
Nice one. Thanks, Q.

General Board / Re: 'Require user lock to edit' and group locking
« on: June 19, 2013, 10:41:53 pm »
Hmmm ... yeah, I see your point, Q.

We have a big model covering disparate aspects of our organisation, and a large user base. The group-locking-at-folder level approach is an option to apply some domain governance, such that a domain (folder) owner could have some confidence that his folder was being updated appropriately (since he controls who is in his group).

However, this approach is onerous and heavy weight and we are also considering simply applying 'Require user lock to edit' and trusting people to communicate appropriately to ensure the right stuff gets done in the right folder (as you suggest).

The Sparx security functionality is quite rich, though, and various patterns are discussed in the help files ..... ... do people tend to use it in some way other than how I've outlined? Or does your point also imply that the default position is that nobody really needs to use any of that function at all?

General Board / 'Require user lock to edit' and group locking
« on: June 19, 2013, 05:57:07 pm »
I've security enabled my project (with 'Require user lock to edit' turned off) and created users within user groups. I've then locked certain folders against certain user groups. This means that the contents of folders can be updated only by those users in the relevant user group.

So far, so good.

I'm now exploring how to have all elements un-editable until a user locks them against themselves to perform edits. As far as I can see, the only way to do this is to turn on 'Require user lock to edit'. However, this means that I can't implement my group locking approach ..... indeed, all group level locks are removed once 'Require user lock to edit' is turned on.

Is there any way that I can allow updates against a folder only for users in a certain group AND make elements in that folder un-editable until the user locks them against themselves (i.e. until the user implements user level locking over and above the group level locking?????

General Board / Disaster recovery (securing my model for restore)
« on: June 19, 2013, 05:40:09 pm »
I need to be able to restore earlier versions of a folder(s), either to roll back changes or in case an entire folder gets inadvertently deleted. Now, while baselining is ideal (for example) to compare versions before restoring, it doesn't help me if the folder has been deleted (since the baselined versions are part of the folder and, therefore, are also deleted).

Is that right? If so, that leaves me with the options of either securing using XMI exports or integrating with a version control system (e.g. SVN). While SVN allows me to restore in a more user friendly way (possibly) it does mean I'm reliant on another system (external to Sparx) and,also, my users have the overhead of installing SVN and maintianing SVN-related user credentials etc.|

Is there anything else related to disaster-recovery that I'm failing to see, or is my only choice between XMI exports/imports and SVN?

General Board / Hijacking somebody else's EA user identifier
« on: June 19, 2013, 05:18:31 pm »
I have security enabled my project, created EA users with the same user idfr as their windows user idfr and turned on 'Accept Windows authentication'. This means, of course, that users can open the project and are "known" to that project by their EA user idfr without having to enter that EA user idfr (i.e. silent login).

So far, so good.

Now I've noticed that any user can then select 'Security/Log in as another user', enter the EA user idfr of somebody else and they are then logged in to the project as that other person. I can't see anything in user permissions that could preclude a user from being able to log in as someone else in this way. One way around this is for EA users to be set up with a password ... but that seems to undermine the benefit of the silent login approach.      

I'm not really expecting anyone to actually do this, but I am being asked the question whether this is a potential loophole.

Any views or ideas?

General Board / Generating sequence diagrams from use cases
« on: May 21, 2013, 05:57:46 pm »
Within a use case I have created a structured specification (within the Rules/Scenarios option).
The available Context References reflect relationships defined elsewhere in the model, of course. When I type in the name of one
of these context references it is "recognized", i.e. shown as underlined.

So far, so good.

Now, when I generate a sequence diagram from here, the context reference elements are created as instances, rather than
links. This means my project browser gets very cluttered, since I'm creating a number of sequence diagrams each of which include the
same context reference elements.
Is there a way to define 'link' as the default representation of elements within a use-case generated sequence diagram?

General Board / Sparx->SVN integration-Windows 7 related problem?
« on: May 07, 2013, 10:21:12 pm »
I used to run v9.3 on Windows XP and integrate with SVN. The first time I did this I had to key my password (i.e. the password related to my organisation - nothing to do with a Sparx EA user id), via a command line prompt, into SVN which cached it for future Sparx->SVN authentication.

Following a alptop upgrade I had to re-install sparx and SVN and I  am now running v10 on windows 7 integrating with SVN (silksvn v1.79). Now, the SVN password won't cache. As a result, I can't set up a folder to be version controlled using Package Control/Version Control Settings. No idea if this is something to do with sparx (v10 specific??) or windows 7, SVN (or a combination of all three)!!!

To make thing more confusing, Sparx doesn't prompt for any sort of password to pass into SVN - it simply fails to set the version control settings and shows an error vaguely related to Password.

Anybody been here before????

General Board / Re: Tracing from one interface to another
« on: May 01, 2013, 08:51:06 pm »
Qwerty - do you mean I create a dependency between component 'Logical1' and 'Physical1', double-click on the dependency arrow and select 'implements' from the stereotype drop down? If so, I don't see 'implements' in the list.

General Board / Tracing from one interface to another
« on: May 01, 2013, 07:40:17 pm »
I've created a component 'Logical1' which exposes interface logicalIF1. This function is physically implemented by 'Physical1' (which has interface physcialIF1). N.B. - the physical interface name needs to reflect a real (version numbered) interface name - therefore, the name of physical and logical interfaces cannot be the same.

Here's my challenge ...... I can obviously realize 'Logical1' component element by 'Physical1' component. But is it possible to link logicalIF1 to physicalIF1 in some way to enable me to trace logicalIF1 is implemented by physicalIF1?  

General Board / Re-using lollypop interfaces
« on: April 24, 2013, 06:36:42 am »
I want to represent a component logically as a single entity (with its 2 interfaces) as well as physically across 2 other components (that implement 1 interface each). As part of that, I'm trying to create those interfaces once and re-use them across all components.
i.e. :- Component1 (the logical component) has 2 lollypop interfaces against it. This is (physically) implemented by Component2, which has 1 of the lolly pop interfaces against, and Component3, which has the other. I can maybe show some sort of link/dependency between Component2/Component3 and Component1 to indicate that the former are physical manifestations of the latter. However, I'm stumped how to make the interfaces re-usable (i.e. create them once, "independently" of the components,and then use them against the component elements as appropriate).
The only way I can see to do it is to create lollypop interfaces on their own and then drag them onto diagrams as links using a realize arrow to articulate the relationship with the between interface and component. However, that doesn't look as succint/tidy on a diagram as lollypops.

Should I be using ports? I've tried, but can't drag/drop an "independent" interface onto the port.

General Board / Merging projects (from baselines?)
« on: March 28, 2013, 12:32:01 am »
I'm exploring baselining function and I understand how to restore eap project X to a basline taken against eap project Y. This, of course, overwrites project X with project Y. Is there a way to merge the two?

General Board / Re: 'Make Design Master' unselectable
« on: March 25, 2013, 06:53:32 pm »
Thanks Simon. That makes perfect sense (and is obvious now you point it out to me!!). At our organisation we're aiming to maintian a persistent view of our (modelled) systems that projects will take a cut of to update and then merge against the persistent view when they've finished. The master-replica appproach seemed to support this. However, we're using a SQL server (to allow sharing etc.). Is there an equivalent approahc, do you know, that might allow us to do this against a SQL DB?????

Pages: 1 [2] 3 4