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 - Van Lepthien

Pages: [1] 2
1
Suggestions and Requests / #Object# macro for Model Search
« on: October 17, 2012, 07:57:32 am »
A number of times I've wanted to use an object's ID when composing SQL searches in Model Search. The latest is "Get all objects whose classifier is X", where X is an object selected in the model.

To be consistent with the use of #Package#, I think it would be appropriate to use #Object#.

I want to be able to do this:

Code: [Select]
SELECT s.Name as ClassifierName, s.Stereotype as ClassifierStereotype, s.Object_type as ClassifierType, s.Object_ID as ClassifierId,
f.ea_guid AS CLASSGUID, f.Object_Type AS CLASSTYPE, f.Name as Name, f.Stereotype
FROM t_object s , t_object f
WHERE
f.Classifier_guid = s.ea_guid
and
s.Object_ID = #Object#

Is this needed enough to make a formal request?

2
General Board / Re: Oracle 11G as repository
« on: January 24, 2012, 06:56:46 am »

3
General Board / Oracle 11G as repository
« on: January 24, 2012, 05:35:42 am »
The EA Systems Requirements page shows the supported Oracle versiona as 9i and 10g.

We are running 11g, and our database support people don't want to run a down-level instance.

We have brought up a relatively small EA repository with Oracle 11g as the database.

We have encountered a problem where attempting to open a package results in a loop - CPU goes to 100% on one of the processors, and it continues to loop until you kill EA.

We are wondering if this is an EA-11g compatability problem.

So:

Has anyone else tried 11g? (I've seen other posts that imply this is so)

Are there known issues? Has the above problem been seen before?

Is there a timeframe for Sparx support for 11g?

Thanks,

Van

4
General Board / Re: EA for the masses
« on: June 30, 2010, 04:37:44 am »
1. The EA model is not consistent with the UML model, nor (my specific headache) is the SysML add-on consistent with the SysML specification. I am having to write my own audit code to check that the diagrams and relationships are consistent with the OMG UML and SysML specs. The mapping between the OMG models and the EA model seems to be kludged rather than well though out.

2. As mentioned by others, the RTF documentation generation leave a lot to be desired. This may be an intrinsic problem of a generalized reporting tool, but the problems are exacerbated because the tool is based on the EA model, while I need reporting that reflects UML and SysML models. The inventive (is that a better word than kludge?) mapping between the models occasionally gives some confusing results in the reports.

3. Lack of visibility of bugs and bug status is especially irksome for those of us responsible for supporting a group of EA users - partially because more users turn up more problems. If there were a bug list available, it would save me a significant amount of the time I take to see if a problem is reproducible (and definable) before I submit a bug report. I suspect that such a list would save a lot of work on the Sparx side, too, by reducing the number of duplicate bug reports submitted. What would also be nice would be a feature that would allow voting on the importance of getting particular bugs fixed.

4. Use of antequated and proprietary technology. The EAP files are in Access 97 format, the Automation Interface  documentation specifies "Microsoft Visual Basic 6.0, Borland Delphi 7.0, Microsoft C# and Java". Thank goodness for Java, but VB6 and Delphi7 have been dead for years. If I need to be writing auditing and reporting programs because of model quirks and RTF reporting inadequacies, I might want to be able to write those in something like Python. If a company bought EA because it was a relatively inexpensive option, it probably doesn't want to shell out the cost of a Visual Studio seat if it doesn't have to.

Van

5
General Board / Re: EA integration with CAC-enabled Subversion han
« on: June 30, 2010, 05:53:51 am »
I have a similar problem when one of my users has a password expiration on the system where subversion runs. As most of them never directly interface with that system, they never see password expiration warnings, which would appear if they were doing command-line svn interactions.

6
General Board / Re: Recover from Version Control without Database
« on: June 30, 2010, 05:30:58 am »
Similarly, you can take baselines and export them and later import the exported baseline.

7
Bugs and Issues / Re: EA - VC desynchronisation / package loss
« on: January 21, 2011, 05:42:22 am »
I suspect the reason "Check out branch" may appear to work is that EA reverts the model content to the content of the SVN XMI file when you do a checkout, and thus the old content pops up. The "Check in branch" brings the database and subversion back into synch if any changes have been made.

This still doesn't explain how things get hosed up in the first place.

8
Bugs and Issues / EA - VC desynchronisation / package loss
« on: May 05, 2010, 03:38:31 am »
We are running EA 7.5 v850, MySQL database, SVN version control. We probably have 10 people working in one model, though usually only 2 or 3 at a time.

There are a couple of peculiar things going on, and I suspect they may be related.

1. One engineer said that a package he had been working in had disappeared from the project browser. I looked in subversion and found the package, and it showd him as having it locked.
At this point I attempted to reload the current project, which had no effect. The XMI file in Subversion was for the package in its initial state, apparently added to VC right after the package was created. He claimed to have checked it in after making changes, but I could find no evidence of that in subversion.
I checked a (full model) baseline that was taken a couple of days before and found the offending package, complete with contents.

2.On what I suspect is a related note, I've found a number of XMI files in SVN that are locked, but either the correspoding packages appear to be checked in (per the browser icon) or cannot be found in the EA project. I have been asked in several instances to help with people not being able to check out packages that they themselves had created - the packages showed as being checked in, but SVN said that the package was checked out. "Resynchronise with VC provider" does nothing to fix the problem - have to go into SVN and clear the lock, after which things usually seem OK.

I suspect a contributing factor is losing connection with the database when EA is open - the people who have this problem seem to all be laptop users, and I, with a desktop, never see anything similar happen when I've been updating things (I also check packages in religiously).

Has anyone else run into similar problems? Are there any suggestion on how to fix them? I've tried telling people to "check in your packages and close EA before you close your laptop", but the problems keep on popping up.

Thanks,

Van

9
Bugs and Issues / Re: DCOM table
« on: September 28, 2010, 01:15:36 am »

10
Bugs and Issues / Re: DCOM table
« on: September 18, 2010, 01:31:16 am »
I can confirm that running EA.exe /register on Windows 7 does not result in an EA.APP entry in the DCOM table (per dcomcnfg.exe).

EA 8.0 build 859

11
Bugs and Issues / VC status problem
« on: June 09, 2010, 08:47:53 am »
EA 7.5 850

MySQL 5.0.37 community server

Subversion

CollabNet Subversion Client 1.6.4

Sequence of events:

Was looking at the ea repository t_package table with the MySQL Query browser, and noticed two checked out packages (as shown in the PackageFlags field)which were not visible in the project browser.

Executed File>Reload current project, and the two packages appeared.

When I right clicked>Package Control>File Properties on one of the packages, it displayed what I expected: the package was locked by the person identified.

When I did the same thing with the second package, I got a popup box

xxx\xxxxxxxx.xml: (Not a versioned resource)
svn: A problem occured; see other errors for details

The t_package table XMLPath specified an existing file out in Subversion, which  indeed was checked out by the person indicated in the PackageFlags.

I executed the Package Control>Re-synch all statuses with VC providers.

When I re-ran the query against the t_package table, both CRC and the lock information disappeared, leaving only the VCCFG parameter.

The packages are still in the same statuses in Subversion, and (to my surprise) I still get the same results when I do Package Control>File Properties for the packages - one shows as being locked, and the other claims to not be a versioned resource.

The only additional factor is that I believe that the user's laptop has gone into sleep mode.

I will be submitting a bug report.

12
The UML 2.1 definition of effect in a Transition:

effect: Behavior[0..1] Specifies an optional behavior to be performed when the transition fires.

So allowing at most one conforms to the UML spec.

13
Bugs and Issues / SysML - how to get dashed controlFlow links
« on: February 02, 2010, 06:37:21 am »
Running 7.5.850, Corporate Edition.

When using the SysML Example Model, I can get a ControlFlow to appear as a dashed line by giving it a <<controlFlow>> stereotype.

In any other model where I create and use SysML ControlFlows (whether or not I have loaded SysML under Add-Ins), I cannot seem to get them to appear as dashed lines. There is no controlFlow stereotype avaliable, and the toolboxes available under MDG Technology for SysML (loades with the sample project) differ significantly from those available otherwise.

What would be nice:

  • The ability to format a control flow as a dotted line. This is consistent with SysML 1.1.
  • An example that was consistent with behavior outside of the example.

Does anyone have a technique for accomplishing the former?

14
Bugs and Issues / Model Validation - ObjectFlow to MergeNode flagged
« on: December 01, 2009, 08:19:33 am »
In a SysML 1.1 Activity Diagram, I have two ObjectFlows going to a Merge node that has a single outgoing ObjectFlow.

When I run Model Verification, I get the message "MRV05002 - error(<anonymous>(ObjectFlow)): ObjectFlow is not legal for Synchronization --> MergeNode".

This is a problem in a couple of respects.

First, the UML specification says:
Constraints
[1] A merge node has one outgoing edge.
[2] The edges coming into and out of a merge node must be either all object flows or all control flows.

Second, a Merge node does not do synchronization. A Join node performs synchronization, a Merge node does not. A Join node with ObjectFlow inputs gets a similar message, but the Join node definition says:
Constraints
[1] A join node has one outgoing edge.
self.outgoing->size() = 1
[2] If a join node has an incoming object flow, it must have an outgoing object flow, otherwise, it must have an outgoing
control flow.
(self.incoming.select( e | e.isTypeOf(ObjectFlow)->notEmpty() implies self.outgoing.isTypeOf(ObjectFlow)) and
(self.incoming.select( e | e.isTypeOf(ObjectFlow)->empty() implies self.outgoing.isTypeOf(ControlFlow))

So either a Merge Node or a Join node may have ObjectFlow inputs, but Model Verification flags them as errors. There are no additional restrictions in the SysML spec that affect this.

(EA version is 7.5.849, Corporate Edition)

I want to get other people to use Model Validation to clean up their models (and I'm considering augmenting the existing validation with add-ins to perform localized verification), but I need to make sure that validation behaves as it should before foisting it on others.

Has anyone else run into similar validation issues?

Van

Addendum: The problem seems to stem from a fork node (one ObjectFlow in, two ObjectFlows out) supplying one of the ObjectFlows to the merge node. I am in the process of checking out whether this is allowed UML.

Addendum 2: Figure 12.73 Control Node Examples (in the UML 2.1.1 Superstructure document) shows a ControlFlow from a Fork Node to a Merge node. Careful examination of the UML spec turned up no constraints that would disallow the same for ObjectFlows.

15
Bugs and Issues / Re: Action as Activity instance?
« on: November 06, 2009, 10:29:47 am »
One thing to note is that in the UML Superstructure, there is no mention of activities within activities. Activities contain ActivityNodes, ActivityEdges, ActivityGroups ("a ... grouping construct for nodes and edges"), ActivityPartitions (a kind of ActivityGroup), StructuredActivityNodes, and Variables.



Activities do not contain other activities. That EA allows one to embed activities is not good UML.

The CallBehaviorAction is how one invokes another activity.

An analogy is to think of activities as being equivalent to callable programs, and a CallBehaviorAction is a call to the Activity (hence the name). If we ignore some better-forgotten programming constructs of the past, it doesn't make much sense for flow to go between a callable program and another callable program without a call.

Interestingly, if you do embed Activities within activities in EA, you get effects analogous to those one might expect with executing callable code inline in some ancient programming language - the constructs that make it OK to use inline make reuse difficult (or, conversely, the constructs that enable the code to be called make its inline use difficult). If you have two activities linked by an edge in one activity diagram, those links persist when you copy the activites to another activity diagram.

It might help to notice that unlike (most) other diagrams, an activity shares the same boundaries as its diagram. One of the problems with the EA implementation is that its easy to create a diagram without the activity.

Pages: [1] 2