Interaction with Visio going forward (post v15.0)

[Paolo F Cantoni]

As I have moved from engagement to engagement, I (as have most Sparx practitioners) have found a plethora of "models" expressed as Visio diagrams.  this latest has proved no different.

However, I  (as most Sparx practitioners) have found it increasingly more difficult to import these diagrams into EA.  Primarily, I believe, because no effort has been put into upgrading the importer.  I tried to import a large diagram recently (a vsdx file created with the latest version of Visio) and there were 1300 (probably spurious) errors reported by Sparx which stopped the creation of any relationships between the items.

Now, elsewhere, both Geert and I (and probably others, but most recently we two) have commented that there is usually a large amount of work required to convert these diagrams into Sparxian.  However, that's not what I'm complaining about this is about actually being able to read the file in the first place.

I'm also not complaining about trying to import a diagram that is visually (i.e. when printed or displayed) "Correct" but under the covers is not syntactically valid (i.e. the connectors don't join the shapes).

As far as I can ascertain, this is a perfectly valid Visio diagram.  But since the Importer hasn't progressed from v1.8.42 for several years, I tried saving the diagram as a Visio 2003-2010 vsd file.  Also to no avail - the same set of errors.

So my question is:
Does Sparx see this as part of its strategic armoury to help users transition to EA or not?

From the Sparx site "Importing Visio Diagrams into Enterprise Architect"

"If you are not using Enterprise Architect to design your systems, you are giving up opportunities for increased productivity, that result from re-using your data across the whole project, sharing that data across the entire project team and using the model to drive the solution."

The importer needs to be updated to allow the better and easier import of Visio diagrams into EA.

Can we have a definitive statement from Sparx on whether they are serious about helping Visio users move their diagrams to EA?

Thoughts (especially from the user base)?
Paolo

[Note: I take Geert's point (made elsewhere) that in many cases the total amount of work to import the diagram using the importer is less than creating the model from scratch by other means.  That having been said, if the judgment call is that it will be easier to import the diagram and post-process the import, then the import SHOULD work!]

[Uffe]

Thoughts (especially from the user base)?

Why yes! I've got some!

I fully agree that functionality as advertised needs to work. This means, among other things, that importers need to keep up with changes in source applications.

I also agree that it is a strategy issue.

Should the Visio importer be just essentially an advertising gimmick?
Or a way to sell consultants?
Or should it be a tool for the end user?

If it's a gimmick, all you need is something that you can advertise. It doesn't actually have to work.
If it's a consultancy hook, it has to work but should require specialized expertise to operate.
If it's an end-user tool, it has to work very smoothly, convert the majority of UMLish drawings with zero hand-cranking, and make sense to a business analyst who is not an EA enthusiast and in fact prefers Visio.

My current client gave up on the Visio importer a couple years back. It was distributed along with the client for general use, but there was no concerted effort to import Visio drawings on masse, more there-if-you-need-it sort of thing.
Departments and projects are not required to use EA or indeed UML, so when people tried it and couldn't get it to work they simply went back to Visio. Some of them did realize EA was the better tool even for their specific needs but they were not prepared to spend the money redrawing everything manually, which is what the level of functionality provided by the importer amounted to.
The lesson there is that when a tool switch is contemplated, people don't accept a very high cost for converting existing materials. The cost of retraining and updating internal processes is high enough.

And by the way, I wasn't being facetious with my consultancy option above. That's a valid option; if you can't (or won't) make the tool end-user-friendly, consultants can step in. But we would then also need support from the manufacturer, with solid in-depth documentation, maybe some transitioning training material, that kind of thing.
(I was being facetious with the advertising gimmick option.)

Anyway. There are some thoughts.


/Uffe

[Richard Freggi]

I think you are very lucky and in the minority to have large quantities of Visio diagrams that actually make sense, first of all individually but even more so as a group.
My experience with Visio is that it's just artwork 90% of the time and needs so much refactoring effort to be used in a real model that building it from scratch in EA is just as quick.  That's just my experience.... so I'm not placing much value on an importer.  (Excel is a different story, it's sometimes the only way to bridge gaps between incompatible tools)

[Eve]

I can only address one point.

But since the Importer hasn't progressed from v1.8.42 for several years

An updated version is included with the MDG Integration for Microsoft Office. Doesn't mean that it's going to do any better on a "model" that only looks valid though.

[Paolo F Cantoni]

I can only address one point.

But since the Importer hasn't progressed from v1.8.42 for several years

An updated version is included with the MDG Integration for Microsoft Office. Doesn't mean that it's going to do any better on a "model" that only looks valid though.

Thanks Eve, I thought that might be the case, but didn't want to try that until confirmed.

Perhaps a link on the old importer page is in order?

Paolo

[Paolo F Cantoni]

Well, I tried an evaluation copy of the Visio Importer in the MS Office MDG.

If anything, it produced a worse result!  The older version picked up some relationships, the newer one (v2.0.1.2) didn't pick up any (at least it's consistently wrong - the earlier one was inconsistently wrong).

So it still begs the question.  Will Sparx support the ingestion of valid Visio diagrams, going forward?
(Eve, I AM making an assumption that the diagram is valid.  Should I send a copy to support for them to analyse why it's not importing correctly?)

I don't mind having to pay for the newer version, but it needs to work!

Paolo

[Eve]

I've personally had little to do with the Visio import. If you're having issues I would definitely recommend sending a sample through to support. At the moment I'd just be making a bad guess.

[Paolo F Cantoni]

I've personally had little to do with the Visio import. If you're having issues I would definitely recommend sending a sample through to support. At the moment I'd just be making a bad guess.

Sample sent.  Interestingly, the import showed NO errors on import (unlike the older version) but still no relationships imported.

Paolo