[cvsnt] Re: Modules and Shared Libraries - tagging strategy?

[Bo Berglund]

On Thu, 11 Nov 2004 22:27:37 +0100, Torsten Martinsen
<torsten at tiscali.dk> wrote:

>Bo Berglund wrote:
>
>>So what we have tried to do is this:
>>In the main project folder we create a batch file we call
>>dependentfiles.bat. This contains a bunch of cvs checkout commands
>>that when run will create a number of folders *below* the main project
>>folder containing the shared code (these are sources, not compiled
>>binaries).
>>
>>When someone checks out the main project he will also get the batch
>>file and the first thing he has to do is execute that. When that is
>>done he has all the files needed for the project inside the main
>>project folder.
>>Note that the batch file is done such that it checks out only those
>>files that are actually needed from the shared modules.
>>
>We also used this approach previously, but have found that using "cvs 
>import" is a superior solution - after all, cvs import was designed for 
>tracking third-party sources.
>

No, you don't understand the situation!!
There are no 3rd party sources involved at all. Instead we have many
projects that use common functionality, which we have stored in a
"Common" repository. We have utility functions and we have common
database classes and things like that.

The real projects are stored in various other repositories, mainly
because of isolation issues when we started CVS. This precludes using
modules and modules2....
So when we have a project module that depends on some common files
that exist in our "Common" repository then we create the batch file to
check out these common files to the project module.
The end result is that we can commit from the project top and we can
tag from there too, and all the files that are involved get the same
tag even if they are on different repositories.


/Bo
(Bo Berglund, developer in Sweden)