Loading ...
Sorry, an error occurred while loading the content.

Who here actually has an integration machine?

Expand Messages
  • Nigel Thorne
    Do any of you guys actually have a separate integration machine, or do you just allow people to check into a shared version control system? The former always
    Message 1 of 65 , Mar 27, 2003
    • 0 Attachment
      Do any of you guys actually have a separate integration machine, or do you just allow people to check into a shared version control system? The former always seemed like overkill to me.

      The cost of checking in is high, so people do it less often... etc.

      cheers
      Nigel
      ----
      There are only 10 types of people in the world. Those that understand binary, and those that don't.



      [Non-text portions of this message have been removed]
    • Bill de hÓra
      ... What was done in C3 sounds like a physical representation of a floating build tag. Here s how it could work in CVS. You: update your sandbox against the
      Message 65 of 65 , Apr 1, 2003
      • 0 Attachment
        Nigel Thorne wrote:
        > This sounds like the ideal setup.
        >
        > Anyone got any idea how to set up something similar in Source Safe or CVS..
        >
        > It sounds like people creating branches, then integrating them back to the main branch... but not quite.


        What was done in C3 sounds like a physical representation of a
        floating build tag. Here's how it could work in CVS. You:

        update your sandbox against the 'build' tag
        do some TDD
        update your sandbox against the 'build' tag
        unlabel the code to check in
        run those tests again
        check stuff into the HEAD tag
        label your new stuff as 'build' in CVS
        everyone else has it when they update against 'build'

        there are synchronization issues and it is a somewhat verbose
        protocol, but you'd doing half these steps in any case (a
        script/task that deals with labelling and relabelling is a
        very-nice-to-have). In practice it becomes a ritual and the
        synchronization issues are about as significant as merging instead
        of locking.

        A pure XP setup might argue that with full test coverage you don't
        need a build tag and can work away against HEAD. In practice, you'll
        often be working against existing code bases and the like that won't
        have full coverage. In non-XP and unstructured working setups, a
        floating build tag is a primary means of staying sane, right up
        there with tests. This is mainly because developers can avoid
        integration paralysis since they won't be breaking a build by the
        mere act of checking in code.

        Bill de hÓra
      Your message has been successfully submitted and would be delivered to recipients shortly.