Topic
14 replies Latest Post - ‏2012-02-25T23:08:17Z by marct22
Dilip.J.Mansukhani
Dilip.J.Mansukhani
98 Posts
ACCEPTED ANSWER

Pinned topic Atomic checkin

‏2010-12-13T15:01:00Z |
With version 7.1 of clear case.The new feature has
been given in this which is atomic check in.
Just want to know who this feature is important
in version management.7.1.1 doc explain that clear case support this with partial implement which means it is not use for ucm.
can anybody explain this feature.

Thanks
Dilip
Updated on 2012-02-25T23:08:17Z at 2012-02-25T23:08:17Z by marct22
  • martina
    martina
    1025 Posts
    ACCEPTED ANSWER

    Re: Atomic checkin

    ‏2010-12-13T15:10:01Z  in response to Dilip.J.Mansukhani
    Atomic checkin is one of those features that CC was lacking in comparison to some other version control tools.
    What it means is that if you do a checkin of multiple elements, all checkins will either succeed or fail. So it implements complete chein or complete failure of checkin for a changeset.

    Welcome to the world of agile/iterative tool development and deliveries. We have it for Base CC now and my guess is that we'll have it for UCM at some point in the future.

    hth

    Martina
    Don't Postpone Joy - Have Fun

    ReleaseTEAM Inc
    http://www.releaseteam.com
    IBM Rational Certified Consultants
    IBM Business Partner
    • SystemAdmin
      SystemAdmin
      47293 Posts
      ACCEPTED ANSWER

      Re: Atomic checkin

      ‏2010-12-13T16:10:05Z  in response to martina
      martina wrote:
      > Atomic checkin is one of those features that CC was lacking in comparison to some other version control tools.

      Of course ClearCase doesn't need this feature, which is dangerous on the contrary! Easy to use for denial of service attacks (against your colleagues, intentionally or not).

      It is just clever wording from subversion marketing to turn a defect into an advantage. Subversion has no support for versioning files separately. It has in fact no support for 'configuration items'.
      In subversion you can only check out and commit directory trees. There are no version trees (you cannot see the branches of a file).
      Subversion cannot decouple checkin from publication.

      To get an atomic publication in ClearCase, apply a label, and rename its type (for instance).

      I really cannot see how atomic commit would make anything more 'iterative' or 'agile'... Bulky and discontinuous, yes. One way to enforce the old main latest syndrome.

      Marc
      • SystemAdmin
        SystemAdmin
        47293 Posts
        ACCEPTED ANSWER

        Re: Atomic checkin

        ‏2010-12-13T16:52:10Z  in response to SystemAdmin
        Atomic check-in is a not required feature in a Configuration Management tool like ClearCase.

        Subversion people, and not only I've add, really believe that "atomic check-in" is an advantage. It is nothing more than a marketing word used to overtrown a weakness.

        martina wrote:
        > Atomic checkin is one of those features that CC was lacking in comparison to some other version control tools.

        ClearCase is not version control, it would saying that a Formula One car is a city car. Sorry Martina.

        Thank God, atomic check-in requires to be enabled in CleaCase!

        BTW I am shocked that IBM implemented this useless feature in ClearCase
      • ChrisMoir
        ChrisMoir
        100 Posts
        ACCEPTED ANSWER

        Re: Atomic checkin

        ‏2010-12-13T16:58:04Z  in response to SystemAdmin
        On the contrary, I can see this being useful especially when dealing with delivers or rebases (in UCM, where this feature isn't yet available). Many are the cases where there is a checkout failing on either operation, and the enduser has proceeded to check in the files that it COULD merge, and then tried to cancel the operation because of the few items that weren't merging -- meaning that I had to go and remove all the checked in versions manually in order to cancel the deliver or rebase.

        Additionally, a failed rebase or deliver wouldn't then potentially block other users while the situation is getting cleared up. On the bad side, it will be harder to troubleshoot if the checkouts aren't still in place.

        I'm not going to argue with the fact that this feature is an attempt to change the subject (ie, turning a negative into a positive) on the part of other source control products (and I say this as a former IBM Rational employee, so take it with a grain of salt) but if customers want it, they want it, and IBM has decided to deliver.

        -Chris Moir
        Inventa Technologies (at Aetna Insurance)
    • Dilip.J.Mansukhani
      Dilip.J.Mansukhani
      98 Posts
      ACCEPTED ANSWER

      Re: Atomic checkin

      ‏2010-12-13T16:13:00Z  in response to martina
      Hi Martina,

      Thanks for this.But can't we do the same thing with clearfsimport.In comparison with the other.
      version management tool i see this is one of the feature which cc was lacking but with 7.1 this as been added.i will be thankful if you can provide me more details on this or if you can point me the link.Lastly if we follow Agility/Iterative which as needing this atomic checkin then it should be in UCM which is RUP base or Agility base.
      Regards
      Dilip
      • SystemAdmin
        SystemAdmin
        47293 Posts
        ACCEPTED ANSWER

        Re: Atomic checkin

        ‏2010-12-13T17:06:44Z  in response to Dilip.J.Mansukhani
        Dilip,

        UCM di per se can be defined AGILE if know how to use it.

        On top of that to cope with multidimensional development infrastructures it can be approach it by AGILE-RUP.

        AGILE-RUP or AUP is a simplified version of the Rational Unified Process (RUP) that describes a simplified approach to developing BA using Agile techniques.
  • brcowan
    brcowan
    719 Posts
    ACCEPTED ANSWER

    Re: Atomic checkin

    ‏2010-12-13T20:37:27Z  in response to Dilip.J.Mansukhani
    Atomic checkins are a study in "be careful what you ask for, lest you receive it."

    In ClearCase, and "atomic checkin" is a series of checkin operations that occur in a single database write transaction. This is both the good and the bad part of it:
    Good: A single checkin failure backs out the lot by aborting the transaction. Simple reversion to the before-checkin state.
    Bad:
    1. In the Database engine ClearCase uses, write transactions are BLOCKING operations, no other operations -- read or write -- can occur while a database write transaction is in process. The "DOS Attack" mentioned above is that you can block all access to the VOB by firing a large checkin operation and not entering a comment for the 2nd checkin. (At least until the transaction times out...) And in any event, even checking in a lot of large files can block access to the VOB.
    2. Checking in to many files in the same atomic checkin operation can cause the transaction files to fill (you would likely have to check in 10's of thousands of files to do this, but some people do this when refreshing 3rd party code drops).

    Issue #1 is most likely the reason this is not supported in UCM environments. Do you really want read access to a PVOB being blocked while a user checks in 50 files? Probably not.

    As long as you're aware of the limitations of atomic checkins in ClearCase, you can decide whether it's a good fit for your process.

    I suspect that Subversion does this not because they want to, but rather because it's needed to prevent users from stomping on each others' work. Who "wins" if 2 users check in changes to the same set of files (remembering that checkouts in SVN default to the equivalent of "unreserved" checkouts in CC) at the same time? Or if you have intermixed changes? Being able to back out your checkin before something breaks becomes critical in a "no reserved checkouts" environment.

    =================================================================
    Brian Cowan
    Advisory Software Engineer
    ClearCase SoftWare Advisory Team (SWAT)
    Rational Software
    IBM Software Group
    550 King St
    Littleton, MA 01460

    Phone: 1.978.899.9471
    Web: http://www.ibm.com/software/rational/support/
    • SystemAdmin
      SystemAdmin
      47293 Posts
      ACCEPTED ANSWER

      Re: Atomic checkin

      ‏2010-12-15T12:58:29Z  in response to brcowan
      We like atomic checkins, we hope it will reduce the number of broken builds on our build servers when developers commit their changes and only partially succeed. Our workaround has been a script to let them know of any upcoming merge conflicts but this state could obviously be changed in the next couple of seconds.

      Atomic checkin is a very old request, I know for a fact that it was raised back in 1994 but then the counter argument from Atria was more about the complexity of multi-db transactions (for multi-VOB atomic checkins) than blocking time.

      Quick question: does the 7.1 Find Checkouts GUI (or the Pending Checkouts GUI in Visual Studio) support the atomic option?

      --Sten Rosendahl
      • bmairpad
        bmairpad
        39 Posts
        ACCEPTED ANSWER

        Re: Atomic checkin

        ‏2011-01-17T10:23:05Z  in response to SystemAdmin
        Any Checkin operation from CCRC Visual Studio .Net Integration perspective will support the atomic option.
        • SystemAdmin
          SystemAdmin
          47293 Posts
          ACCEPTED ANSWER

          Re: Atomic checkin

          ‏2011-03-07T08:32:51Z  in response to bmairpad
          Does it support it silently or do you have an option as on the command line? What about the Find Checkouts GUI?
          • bmairpad
            bmairpad
            39 Posts
            ACCEPTED ANSWER

            Re: Atomic checkin

            ‏2011-03-07T12:34:37Z  in response to SystemAdmin
            Atomic checkin is not supported silently. User needs to explicitly mention that.
            User can do so, by checking the option "Commit changes in one transaction per vob" in CCRC Checkin GUI. (Given that the vob is enabled to handle atomic checkin)
            Also atomic checkin is not possible from Windows Native UI.
            • GKellner
              GKellner
              258 Posts
              ACCEPTED ANSWER

              Re: Atomic checkin

              ‏2011-03-07T15:39:18Z  in response to bmairpad
              The support for all GUIs regarding the atomic checkin will be part of ClearCase 8.0.
              This is the info I got from IBM support.

              greetings georg.
              • marct22
                marct22
                488 Posts
                ACCEPTED ANSWER

                Re: Atomic checkin

                ‏2012-02-25T23:00:06Z  in response to GKellner
                Just an FYI, I don't see any non-CTE GUI that can support atomic checkin options

                I don't see it in "Find Checkouts", ClearCase Explorer, not in version tree. If I right-click on any file and checkin, it's the standard questions (use orig comments, identical, preserve), and if i check-in, it behaves as normal. No atomic on the GUIs (other than CTE), so if you are using snapshot views, you are out of luck with CC 8.0.0.1
                Marc Towersap
                Sr Consultant
                ReleaseTEAM
                mtowersap@releaseteam.com
                • marct22
                  marct22
                  488 Posts
                  ACCEPTED ANSWER

                  Re: Atomic checkin

                  ‏2012-02-25T23:08:17Z  in response to marct22
                  By out-of-luck for snapshot, I mean you have no GUI options for atomic commits. ct ci -atomic

                  Marc
                  Sr. Consultant
                  ReleaseTEAM
                  mtowersap@releaseteam.com