On Wed, 17 Jul 2002, Andreas Otte wrote: [ lines reformatted to 79 columns ]
> Hello! > > I'm evaluating using CVS for our software development as VC-tool. > > Here is the scenario: > > We have a product in constant development with intermittend > releases. Usually a customer gets the product in a certain version > (product release) and also gets specific changes to some files of that > version (only 2 or 3 percent). These changes are specific for that > customer and we want to track them. Such changes are not really a parallel stream of development; it's not appropriate to use CVS branches for this type of thing. Use some other configuration mechanism to represent these variations in a single code stream. This is similar to targetting multiple platforms. If you wanted each release of some program to run on half a dozen platforms, you wouldn't create half a dozen branches. Rather, you would make use of tools like conditional compilation mechanisms built into your programming language. > So I'm branching for that specific customer at the relaese-point and Ask yourself whether this is a process you want to repeat each time you release. >create a new working copy with cvs co -r CUSTOMER_BRANCH module. Then I >do the specific changes for the customer on that branch and tag my real Isn't it too late to be making changes? You have a release point, which means that the normal configuration of the software, as well as the customer-specific configuration, should have been already put through testing. _______________________________________________ Info-cvs mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/info-cvs
