branching problem

[doubleK]

Hi,

I tried a couple of things concerning "continous integration" and "brunch by feature" (branch by task)  topics.

But I have some problems to get it run in the way I would like to.

 

I think about something like this:

 

==> unfortun, the garph is removed ==> attachment

 

I would like to have a main-line which represents the source code.

From the main line I create serveral braches (Feature-1 ,.....). From each of this branches I would like to have a test-branch (TB) which includes the test code for the feature (which should be a working project as well)

 

We work with the IAR Workbench (compiler, linker, .......included). So in order to have a project, we need a IAR Workspace. So there are a couple of configuration and setup files which belong to project setup, but not to the source code (*.ewp, *.epp, *.ewd, *.dep, *.dbgdt, *.dni, *.wsdt, ...)

 

Another basic-requirement for me is, that all the branches should be a running project in it self.

 

After developing the Feature-1 (compiling without errors) the source for Feature-1 should go into the Test-branch for testing. If everything works fine the source-code should be merged from /main/Feature-1 into the main-line. If there are problems during tests, the development on /main/Feature-1 continous, the new changes should be merged back to the Test-branch, tested again.......and so on. (not displayed in the graf,)

 

My big problem are the configuration and setup files from the IDE (IAR Studio).

The files are, of course, different for /main/Feature-1 and /main/Feature-1/TB, and therefore plastic tries to merge them as well.

 

Another problem is, that in /main/Feature-1 there are conf/setup files as well, which will go to the main-line after the merge as well (added items.....)

 

I hope someone understands what I would like to express.

 

At the moment I think I bark up the wrong tree.

How do other handle there projects ?

 

May be it´s to much what I would like to have ?

 

 

Any ideas are appreciated .....

 

Best regards

Karsten

 

 

[doubleK]

Hi,

is it really so unclear what I mean ?

 

Best regards

Karsten

[SilverKnight]

You would probably be better off to make a new test project for the test branch with different names - however, your IDE may not allow that, as these may be specific IDE settings if I understand correctly.

 

If that is the case, then it seems to me that you would have to ALWAYS do cherrypick merges where you UNSELECT the project files (*.ewp, *.epp, *.ewd, *.dep, *.dbgdt, *.dni, *.wsdt, ...) from the merge operation - since you actually want different projects with the same name in the same repository on a different branch.  I expect that would have to be done for any merge, weather to the test branch, or from the test branch.