Again about repository

[JV]

Is there some white paper how to use repository?
Yes, I know how to setup it and how to get project from repository but from where I could get answer, How to add several projects into repository and how to get tem from repository?
Currently I can not do it, existing project in repository is overwritten by new added project.

thank you!

[«Midnight»]

JV,

You need to create a new schama (or equivalent) repository for each project. You can have multiple project roots in a repository, but I don't think that is your current problem so I won't deal with it here.

Depending on the type of repository you use, you need to go about this differently. For example, if you use SQL Server, you need to create a new database, with a unique name, etc. This is the data source you will select if you connect through the OLEDB or ODBC connection dialogs. If you were using Oracle, you would create a new user, which gives you the same results. You might want this to be a 'dummy' user, and then assign access rights to other users so they can connect. Finally, if you want to connect to an EAP file through the repository interface (there are some reasons for doing this) you would use a separate file for each model, much as you would if you were just opening files through the desktop interface.

David

[JV]

thank you.

yes, I use oracle for repository. So, we have about 100 projects what are under development at the same time. According to this I need at least 100 schemas (if i remove project from repository when it is completed).

But will be there in some day possibility to use same schema for multiple projects?

[«Midnight»]

We live in hope JV. That's not what you wanted to hear, I know, but it is where we are now.

But all is not lost! If you've got an Oracle set-up of that size you have also doubtless got a few Oracle wizards around, trying to fill the empty hours now that everything is working so well...

Have one of them create a wrapper for the Sparx procedure to set up a project. The wrapper should be able to take either a single project name or a set - I'll leave you and your wizard to figure out the most effective way to pass parameters. I would then create the back end, probably giving it an 'internal' name geared to EA models, run the Sparx script to set it up, and then provide whatever default user access you might define.

Once this works well, have your wizard look at how EA does its data transfer function to set up a project. Take the before and after pictures of a new, empty project in Oracle. Then have your wizard build an ETL script that does the same thing, and can be called from the above model creation script. Try setting this up with an option to import a specific file containing reference data, to allow you to set up certain classes of models with corporate standards (or whatever) reflected.

At this point you will have a tool that allows you, or your DBA, to build a new model schema on the fly, populate it with core data, and set up basic access.

Yes, it is some work, but this is something that Sparx could not do for you (we all claim our business are unique, but I think they might all be identical with the exception of the Oracle set up). It should pay off handsomely long before you reach the 100^th model!

HTH,
David

PS: If you feel like giving something back, you could always publish the script (after obtaining permission of course) to the EA User Group and Wiki - see the Automation section of the forum for details.