DBMS Repository Methodology

[mwtb]

We've recently upgraded to the Corporate Edition in order to get more robust sharing of EA projects and I am in the process of setting it up and creating the documentation on how to use it.

I have a MySQL database up and the connection is working fine, but I am not entirely clear as to the best way to approach the project structure issues. I'm hoping someone can give me the benefit of their experience.

Because we have been using EA with EAP files and had issues when trying to share them, we have gotten into a (possibly bad) habit of creating our own projects for specific pieces of work in order to limit the numbers accessing any one project file. Now, here's where I may have a misconception that someone can set straight, but as far as I can tell, if I want to create a DBMS repository for a project, I need to actually create a new MySQL database  and this has to be done from outside of the EA environment. Am I missing something here?

Obviously, we are trying to work out of the same project as much as possible, but we are bound to need to create separate projects quite often. How do other people deal with issues like codebase branching and merging, for example.

Feel free to just point me at some documentation if it exists, but I have been digging around and even the EA User Guide has the entire section on just setting up and using a single DB repo missing (the links go nowhere).

[thomaskilian]

Hi,
there's a lot of disussion here about using a repository shared for several people. I guess this is a candidate for the Sparx FAQ (to be continued...). Please try searching for "Repository", "Project Management", "Sharing" a.s.o. (increase search time to more than 7!). I can't tell much about since I'm more or less a single user

You are right that you have to setup the repository from outside. You can download the scripts from EA's registered user page. AFAIK there's also one for MySQL.

[dmag]

You can also read the following article :

http://www.sparxsystems.com.au/downloads/whitepapers/EA_Deployment.pdf

(this link is from the build 769. The one in the
766 doesn't work any more).

I'm currently trying to find the best way to share models, I'm looking around repository and version control in Clearcase. If some one has any experience...

[thomaskilian]

Try looking for "Clearcase" in this forum

[mwtb]

 I have searched on those terms and set the search limit to several hundred days (why is the default 7?). The problem with search is that you can't necessarily tell if you've covered the ground thoroughly enough.

Thanks for the pointers to the DB set-up page, but, as I said, I have the database and an ODBC connection set-up. I'm not looking for information about that, outside of one question, which is related to my general surprise at how clunky the multiple repo handling is from a tool perspective:

I have an ODBC connection that declares itself to be working via its own test whether I set a default database or not. In Data Transfer, if I have set the default db in the ODBC connection and I just leave the EA connection details blank, it works and I can export data to that DB.

With the same setup, I can also see a list of the databases on the server as "catalog"s, however, if I select one of them, even the database defined in the ODBC connector, it fails (catastrophically, apparently).

If I leave the ODBC connection database entry blank, EA can't see any catalogs, nor will it connect if I type the catalog in.

As for an FAQ entry, I think the user manual would be the place for this sort of basic "get it working" info, especially for functionality that is only part of the premium product line. This is all a bit of a pity, because I'm a big fan of the core product.

[dmag]

Are you using the Microsoft ODBC provider ?
When I set-up my repository in a Oracle Database, I have the same trouble (when I test the connection, it works, when I try to connect, it doesn't work). It's working since I use the Oracle ODBC provider. I don't know if there is several provider for MySQL, but you can investigate this way.

[mwtb]

Are you using the Microsoft ODBC provider ?

No, I'm using the ODBC driver from MySQL themselves. I'm not aware of any alternatives.

[thomaskilian]

...
 As for an FAQ entry, I think the user manual would be the place for this sort of basic "get it working" info, especially for functionality that is only part of the premium product line. This is all a bit of a pity, because I'm a big fan of the core product.

You're probably right. The tool has obviously some "space" for improvements especially regarding documentation.

I normally work with EAP files but I tried with ODBC also. I remember that this worked fine. Also I remember that the ODBC provider is important (I used MS with SQL Server).  So probably you should investigate there.

[mwtb]

It's not a case of it not working at all, it's a case of it working in a manner that is less than easy for the users. I was hoping that someone would have experience of using EA with MySQL and be able to tell me that what I am seeing isn't how it is intended to work.

To simplify, what I appear to have is a scenario whereby for every new EA project I will have to:

1/ Create a new database.

2/ Create a new ODBC connector on every machine that will need to access that database.

The first point is annoying but livable, the second is entirely unreasonable and I can't believe that's how it is intended to function.

[thomaskilian]

What shall I say? That's ODBC, not EA. Except you have some remote control tool, you will have to live with that.

[mwtb]

What makes you say it is ODBC? Other ODBC apps seem capable of selecting a database from a server based on a single connector and the EA interface itself suggests that they intend it to work that way, so why are you saying that my problem is with ODBC?

I appreciate from your responses that you feel the need to defend EA, but I'm not here to attack it, I'm genuinely surprised that this doesn't seem to work properly.

[thomaskilian]

Well, you're right that I like EA, but I also know that it has its deficiencies.  However, in this case I think it's ODBC, not EA. If, for example, I set up an ODBC connection to MS SQL it is completely different to an Access connection - or even MySQL. For SQL Server you have to supply identification information (which usually is no fun) and with Access you don't.  If you are introducing a new repository, you have to create an ODBC connection on each machine separately, supplying all the neccessary information. I don't know of any ODBC-aware application to do that for you.

Or maybe there's some misunderstanding? Do you want EA to talk natively to the database?

[mwtb]

 Well, for anyone interested in a workaround, and in the absence of anyone telling me of a better way, here we go:

Here is a step-by-step to allowing an EA client to access multiple databases without having to add unique ODBC connectors.

This assumes that you have a single MySQL server holding your EA repos and that you've installed the ODBC driver and have a repo database or two already setup. It is based on Windows 2000 with the latest 3.51 MySQL ODBC Driver from www.mysql.com. I'm also ignoring the user/password parts as my needs in this area are probably different from most.

* Create a DSN and related MySQL ODBC connector for your server, and set the catalog to be any valid database on your repo-server.

* When you come to import or export to a database in EA, select the MS OLE Provider for ODBC Drivers.

* Click Next.

* Select "Use Data Source Name" and choose the DSN you setup previously. You can now see the names of the databases in the drop-down list, handy for forgetful people or for generally having a system that is halfway usable. Do not select one of the names, it won't work.

* Now you have your target database name, select "Use Connection String" then click "Build".

* Select your DSN from before, click OK.

* Selecting a different database in the connector config won't work, so click OK again.

* Now, edit the database name in the connection string to match your target repo.

* Click OK.

* Now you have to select the DSN again for some reason and click OK.

* Give it a name if you're connecting to view/use the repo.

* Cross your fingers.

If anyone has a better way, I'd be happy to hear about it.

[thomaskilian]

You're right I just installed MySQL here and tested it. Awful

I recommend to send a bug report to Sparx. The not working steps are definitely bugs.

Edit: During transfer of t_stereotype I received an abnormal termination

[mwtb]

It's not a bug with EA as much as a bug with EA's documentation. The issues surrounding setting up the catalog look like issues with the MySQL ODBC driver (not problems with ODBC itself).

I'm hoping that we have a spare SQL Server or Informix licence hanging around so I can try with those and see if it's a smoother process. In the meantime I need to start looking at version control, which from a scan of the documentation looks equally poorly described from the perspective of someone who wants to understand how their usage of the tool might be structured.