TAM - Incorrect item cloned!

[Paolo F Cantoni]

We've found what we consider a SERIOUS defect in TAM (Time Aware Modelling).

Scenario:  We have a set of current items, in diagrams in a branch versioned "Curr".

We intend to make a RoadMap branch indicating the "To Be" state.  We clone the Current Branch - which gives us new Folder structure and Diagrams (but not the items) - which is what we want; to a branch versioned "RdMp".  Since we only want to clone items that will change in the future.  We clone an item in a RdMp diagram to version "RdMp".  The item is cloned correctly; so far, so good.

As part of our on-going repository management, every year (at the end of the year) we intend to clone Current into a "Past" "snapshot" that describes the state at the end of that year; giving us the ability to "compare and contrast" over time.  So I select the Current branch and say Clone structure (in this case, to a folder structure versioned "Past").  Again, folders and diagrams cloned, no items, as desired.  This time, however, to achieve our goal of snapshotting the entire Current branch we have to clone ALL the items in the "Curr" version.  We select all the items in the diagram in "Past" diagram (which are all namespaced "Curr::") and say Clone to new version (to "Past").

Imagine our surprise when for the item in "Curr" that had been cloned to "RdMp", instead of getting a new item in "Past", we got the "RdMp" item on the "Past" diagram! 

We confirmed that the original item on the Current diagram was still the "Curr" item and that that item WAS NOT CLONED into Past as a new item.  This is a defect, the original "Curr" item (on the diagram) should have been cloned, not the "RdMp" item (which was nowhere to be seen).  We also confirmed it's not a caching problem since we exited Sparx and re-entered and then cloned the structure. This needs to be fixed urgently!

Reported,
Paolo

[Paolo F Cantoni]

The initial response from Sparx is that they can't reproduce the problem.  No real surprise there.  They asked for a sample repository, which we're preparing.

In doing so, the problem seems to get "worse".  We exported the test branches and imported into a clean repository.  So, at this stage, one item was cloned to "RdMp" and the other 3 were cloned to "Past".  So, to check if the problem is still present in the "clean" repository, I go back to "Curr", clone structure as a new version to "Pst2".  The diagram (in "Pst2") shows all items from "Curr".  Select all items in the diagram, clone to new version "Pst2", the end result: 3 items from "Past" and 1 from "RdMp".  So it seems to get confused as to which is the master and which is the clone.

Paolo

[Paolo F Cantoni]

Sparx replied more formally, once they were able to see the issue:

Our developers advised that this is the intended behavior. Time-aware versions can be built sequentially, but not in parallel. You can find the sequential relationships in the traceability window.

As far as we're concerned, We need to be able to clone in parallel.  You have multiple options which you need to model simultaneously.  Only one (or none) of which will "get the guernsey".  we're still investigating if we can do it by enhancing the standard functionality.

Paolo