User Security and Version Control

[ssands]

Hi,

I am unclear about what appears to me to be overlapping functionality between User Security and Version Control.

Will my modelers have to apply two locks to edit a diagram? (One from User Security and one from Version Control)?

I would greatly appreciate hearing from people that have implemented both features and best practices they have learned.

Thanks in advance!

[qwerty]

You should use either the one or the other, not both. If you use VC a lock will be applied so others can't edit.

I always vote for using locking via RULtE (Require User Lock to Edit) and make snapshots on a nightly basis to check into VC (a script can do that automatically). VC should only be used in a very distribute environment where people share the model via local EAP files. If you can use a central database then use RULtE.

q.

[ssands]

Thanks for your reply.

We are already using a database as our repository and SVN as version control. I don't see a compelling reason to move from that. I also don't really want to have to create and then run a  script to put changes to VC each night.

I was looking to User Security to place control over other aspects of Sparx interaction.

Thanks again.

[qwerty]

If you use "normal" security/locks then you won't have any issue (except some of your modelers tend to go wild and set additional locks). Just don't use the locks and be happy with the other privilege methods.

q.

[Geert Bellekens]

You should use either the one or the other, not both. If you use VC a lock will be applied so others can't edit.
q.

Actually you can (and we do it, using "Require user lock to edit" mode).

When checking out a package EA will automatically propose to lock the packages as well.
Consequently when checking-in EA will propose to unlock the package.

This mechanism is usually enough to keep both types of locks in synch, but you can still apply/release user locks seperately.

The important thing is that you need both lock (VC lock and user lock) to be able to edit.

The reason we use both is that we use VC only for a small shared part of the model.
All the rest is not controlled by VC.

Geert

[ssands]

Thanks, Geert.

We use VC on the whole model, but I also want to allow other users to browse, but not change. Also, I want to be able to set up certain members of the Arch team with higher privledges than others. I'm trying to find the right mix of features that will let me do that.
Since our repository is a database, I'm actually thinking of making a file-based copy for read-only, so I don't have these read-only users accessing the database.
Still working out the details....

[AndyJ]

For read-only access, there are a couple of other methods available.

You can distribute EA-Lite which is view only, or publish your model to HTML on a web server and distribute links. (the HTML model is the easiest method from the point of view of the read-only user) Especially as you can embed static links into confluence pages (or web pages, or sharepoint or whatever...)

I've used both methods.  HTML model for external customers and EA-Lite for SMEs etc. who are located within the team.

[Geert Bellekens]

For readonly access you can choose for

- EA Lite
- DB user security
- EA user security
- VC user security

I wouldn't bother with creating a file copy, because then you'll need to keep that updated etc...
One of the above options is much less hassle.

Geert

[ssands]

I might have to use DB and VC security. Some of my users will have full versions of EA, so EA-Lite wouldn't work for them. And EA User Security too!

[minime]

Hi all,
We use the both locks and "The Require User Lock mode", which leads to a curious error message when you try to check-in a "red balls" package (This package is version controlled, but you checked it out whilst not connected to the version control server.). I've got a misterious error message telling that

the current action cannot be completed as the necessary security is not available

With help of Google and an hour of my time I was able to guess that the message might have something to do with the Set Security Policy help topic and deduced that whilst a check-in is essentially a change, I might need to lock the package first.

Yet there is another help topic, Offline Version Control / Checking In a Package That Was Checked Out Offline. Of course it does not mention neither the vague error message nor the need to apply user lock before this sort of check-in. I can see in my crystall ball that the two threads of functionality were implemented by two teams of developers, testers, and techwriters not talking to each other. And there is no decent testing process for User Guide in Sparx. Please apply a bit more effort in this direction.