Dave Belanger, David Korn, and Herman Rao. Infrastructure for wide-area software development. In Proceedings of the 6th International Workshop on Software Configuration Management, pages 154--165, 1996. Available as volume 1167 in Lecture Notes in Computer Science, Springer-Verlag.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....and convenient for users, we designed it as an extended file system. All versions of all source code stored in the repository are visible in the file system name space, so users can browse and compare them using existing, familiar tools. We do not provide mechanisms such as views [2] or viewpaths [3] that can make the meanings of names appear to change depending on the user, the environment, or what versions currently exist. The repository plugs into the file system name space by acting as an NFS [19] server. It also exports a separate remote procedure call (RPC) interface, for access to ....

....(such as compilers and linkers) can be run, called volatile directories. Volatile directories enable the builder to run build tools in an encapsulated environment, in which only the desired versions of sources and derived files are available. Thus they give at least the same power as viewpaths [3], but under direct control by the builder. Volatile directories also let the builder detect the precise dependencies and results of each build tool invocation: the repository tells the builder exactly which of the available files and directories the build tool accessed and what new files and ....

Dave Belanger, David Korn, and Herman Rao. Infrastructure for wide-area software development. In Proceedings of the 6th International Workshop on Software Configuration Management, pages 154--165, 1996. Available as volume 1167 in Lecture Notes in Computer Science, Springer-Verlag.

....CM systems can be implemented based on the abstraction layer defined in this paper, while exhibiting the same qualities as a CM system built on top of a traditional database. NUCM is only one of several systems that fall into the category of Figure 13c. The other systems are CME [25] Gradient [4], CoMa [51] and ScmEngine [10] CME extends RCE [26] itself a programmatic interface to RCS [47] with collection management. CME is similar to NUCM in that it provides an architectural separation of the repository from the actual system that stores and versions the artifacts. However, two ....

D. Belanger, D. Korn, and H. Rao. Infrastructure for wide-area software development. In Proceedings of the Sixth International Workshop on Software Configuration Management,number 1167 in Lecture Notes in Computer Science, pages 154--165, New York, New York, 1996. Springer-Verlag.

....needed during the development of a software product, configuration management tools were among the first faced with the reality of having to operate in a distributed setting. In response, many distributed CM systems have been developed in the past few years (e.g. ClearCase MultiSite [1] Gradient [3], ScmEngine [5] DISC [12] WWCM [13] DSCS [16] Perforce [17] Despite this broad availability, it is clear that a single system has emerged as the de facto configuration management system used in Open Source projects. In fact, this CM system, CVS [4] has been adopted by almost every major ....

D. Belanger, D. Korn, and H. Rao. Infrastructure for wide-area software development. In Proceedings of the Sixth International Workshop on Software Configuration Management, number 1167 in Lecture Notes in Computer Science, pages 154---165, New York, New York, 1996. Springer-Verlag.

....to do with project structure than configuration management, must eventually be merged by the users themselves into a single baseline version. This can be a serious burden, especially if the attributes and relationships on the artifacts have changed as well. Gradient Similar to ClearCase, Gradient [3] is a CM system that is based on replication. But, unlike ClearCase, replication in Gradient is automatic and continuous. Each update that is made to an artifact is broadcasted instantly as a delta to all replicas. Because Gradient only allows incremental modifications 3 to the artifacts it ....

D. Belanger, D. Korn, and H. Rao. Infrastructure for Wide-Area Software Development. In Proceedings of the Sixth International Workshop on Software Configuration Management, number 1167 in Lecture Notes in Computer Science, pages 154--165, New York, 1996. Springer-Verlag.

....Below, each core policy is mapped onto the abstraction layer and discussed in detail. 5.1.1 Checkout Checkin The rst two CM systems, SCCS [Roc75] and RCS [Tic85] pioneered the checkout checkin policy. It has since been the basis for numerous CM systems, including DSCS [Mil97] Gradient [BKR96] Ode [ABGS91] Sablime [Bel97] ScmEngine [CPT97] and SourceSafe [Mic97] The policy focuses on providing version support for individual artifacts and is typically based on the combined use of a repository and a le system. Versions of artifacts are stored in the repository, but users do not ....

....of Figure 5.3 to perform concurrency control is now super uous: the version tree artifact is already locked and the procedure checkin does not need to perform any additional locking. The creation of a branch is disallowed. In some policies, it is required that linear evolution takes place [BKR96] In these policies, branches cannot be created. Two modi cations, each leading 85 to a slightly di erent behavior, can be made to the standard checkout checkin policy. In the rst, the procedure checkout is modi ed to, a priori, prevent the creation of branches in the procedure checkin. With ....

[Article contains additional citation context not shown here]

D. Belanger, D. Korn, and H. Rao. Infrastructure for wide-area software development. In Proceedings of the Sixth International Workshop on Software Conguration Management, number 1167 in Lecture Notes in Computer Science, pages 154-165, New York, New York, 1996. SpringerVerlag. 214

....collaborative authoring of the same HTML file. Since VTML stores all changes to a file, no locking is necessary as the overwrite problem cannot occur because no data is ever overwritten. 4. 2 Distributed Configuration Management A common problem faced by Delta V, ClearCase [22] and by n DFS [3], is the desire to improve a pre existing system by adding CM support. Both ClearCase and n DFS add this support to the filesystem, and hence the low level interface to data stored in the system is via operating system library calls. In contrast, since Delta V is building upon the Web, access to ....

D. Belanger, D. Korn, H. Rao, "Infrastructure for Wide-Area Software Development" In I. Sommerville (ed.), Proc. SCM-6, Software Configuration Management: Selected Papers, LNCS 1167, Springer-Verlag, ICSE'96, SCM-6, Berlin, Germany, March 25-26, 1996, pages 154-165.

....to improve end to end network latency. The idea of trading off computation for I O bandwidth has appeared numerous times in past systems. Examples include application specific deltas and compression, such as Low bandwidth X[15] compressed network or disk I O [3, 6, 7] replicated file systems [2]; shared memory [16] and checkpointing [9, 19] It seems that the same tradeoffs apply in the domain of the W 3 . We address the issue of latency from the perspective of sending the differences between versions of a page, or deltas, in order to avoid sending entire pages. Briefly, deltas are ....

Dave Belanger, David Korn, and Herman Rao. Infrastructure for wide-area software development. In Proceedings of Sixth International Workshop on Software Configuration Management, March 1996.

....to do with project structure than configuration management, must eventually be merged by the users themselves into a single baseline version. This can be a serious burden, especially if the attributes and relationships on the artifacts have changed as well. Gradient Similar to ClearCase, Gradient [3] is a CM system that is based on replication. But, unlike ClearCase, replication in Gradient is automatic and continuous. Each update that is made to an artifact is broadcasted instantly as a delta to all replicas. Because Gradient only allows incremental modifications to the artifacts it manages, ....

D. Belanger, D. Korn, and H. Rao. Infrastructure for Wide-Area Software Development. In Proceedings of the Sixth International Workshop on Software Configuration Management, number 1167 in Lecture Notes in Computer Science, pages 154--165, New York, New York, 1996. Springer-Verlag.

....developing a superior software product. It may also be possible, if project members are on different continents, to schedule work hours such that the project is being worked on around the clock; even though all project members are working in the normal working hours for their respective time zones[3]. Research into this area has resulted in the development of a number of environments that partially address the goal of wide area software development. Some currently implemented systems that aim to coordinate groups of distributed software developers include WebMake[2] OzWeb[10] Alliance[5] ....

....area has resulted in the development of a number of environments that partially address the goal of wide area software development. Some currently implemented systems that aim to coordinate groups of distributed software developers include WebMake[2] OzWeb[10] Alliance[5] Share[18] and GRADIENT[3]. These implementations range in functionality from providing configuration management for distributed artefact repositories through to the facilitating the definition and enactment of distributed software process descriptions. 1.3 Thesis outline Chapter 2 of the thesis gives a more detailed ....

D. Belanger, D. Korn, and H. Rao. Infrastructure for wide-area software development. In Software Configuration Management: ICSE'96 SCM-6 Workshop, Berlin, Germany, May 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC