RDBMS and VC?

[Craig-Mitchell]

Have you verified that the 32bit PostgreSQL ODBC drivers are on the path when you start EA?

[robert.staflin]

Have you verified that the 32bit PostgreSQL ODBC drivers are on the path when you start EA?

Yep, now I have added it both to the PATH and CLASSPATH (the latter is useless, of course). No change.

I have managed to transfer the project without a hitch to an SQL Server Express on the same machine, so maybe I'll stick with that for now. Sigh - I'd really like to get rid of that SQL Server!

[ChrisBD]

I have found that the EA error messages here can be a little misleading or at least not very helpful.

When you setup your database did you use the scripts supplied on this site to initialise it? EA expects certain tables to be present within the empty database before it can transfer the project.

Also is your datasource set as a SystemDSN and does it point to the initialised database? The ODBC connection test will return an okay result even if youdon't have a database schema selected.

[robert.staflin]

Yes to all! I tried using both a SystemDSN and the other one. It seems Sparx support has given up - they asked if I've set up the ODBC conf as described, which I said I had - and they haven't followed up since April 16...  

[ChrisBD]

Have you had a look into the windows Event Viewer to see if there is any light to be shed?
Do you have access to the server logs?

[RoyC]

Robert

Support have not given up on you, they are just trying to get the right machine and system to run tests on. You should get further responses in due course.

[robert.staflin]

Yes (or actuall no), I tried again and have looked through the Event Viewer at the logs. Nothing.

[robert.staflin]

Robert

Support have not given up on you, they are just trying to get the right machine and system to run tests on. You should get further responses in due course.

Great, thanks!

[robert.staflin]

Robert,

It's not the eap file that is VC'd, but the xmi export file.

Geert

OK, I've been stuck for a good while on this again. It's the "hen and egg" problem. I've sucessfully transferred to an SQL Server DB now, and have also sucessfully done VC. Together though...

Should I
1. Make the .eap file, import that into the svn repository, check it out and then transfer that to the database?
2. Transfer the .eap to the DB, then export something to XMI, import that into svn with svnimport and then check out the project?

[Geert Bellekens]

Robert,

You transfer the eap file to DB.
Then you setup VC integration in EA
Then you use the build-in EA functions to control the packages in the VC.
From that point on you use the EA functions to check-out and check-in the controlled packages.
The fact that EA uses xmi in the background is in fact completely transparent, in theory you should not even be aware of that technical detail.
Once the VC integration is set up you should never do anything directly with the VC, but only through EA.

Geert

[robert.staflin]

Great - I think I've got it working now. One last thing though (I hope ;-) ): Do I have to tell EA to version control all the sub packages?

At the moment I've marked the entire tree for VC, and EA shows me that its' sub packages are then VC:d. The packages "below" that aren't.

So - do I have to explicitly tell EA that I wan't to VC those? When the top level was marked for VC, it looked as if all the "leaves" were traversed too! Maybe all the packages that someone might want to check out separately need to be explicitly added for VC?

Thanks in advance.

/Robert

[Geert Bellekens]

Maybe all the packages that someone might want to check out separately need to be explicitly added for VC?

Indeed.
You could have EA do that for you using the "Add Branch to Version Control". That will add each sub-package separately to VC, but the downside of that function is that it uses the ID of the packages as filename, which makes messing with the VC manually a bit complexer.

Geert

[robert.staflin]

OK, it seems I'm forced to ask one more question. It think I read somwhere that I had to configure VC to enable anonymous access, but in the Version Control pdf on Subversion, it just says that there should be user names and passwords setup.

So - where do I configure the user name and password that EA authenticates as? At the moment, I can modify the repository only from the server - and I haven't done any user authentication setup at all in the svn config file (if I do, I get errors that I'll have to deal with eventually). Is this done like with CVS?

[robert.staflin]

Solved, I think. I just needed to work a little bit more on my Subversion skills, and also check out again after I'd made the server authentication setting changes (to get the client to authenticate once).