Subversion 1.8 Provides Automatic Merge and Significant Storage Savings

Subversion 1.8 Features Simplify and Automate Merge and Reduce Complexity

While Apache Subversion simplifies development, helps with team synchronization and offers advantages in terms of handling large files and repositories, it has been limited in its merge capabilities. This has led some developers to move to other SCM systems. With the long-awaited release of Apache Subversion 1.8, users can look forward to more complete merge functionality as well as storage savings.

In the past when using branches, users needed to determine if they wanted to do a sync or a reintegrate merge based on the direction of the update between branches. With Subversion 1.8, merges are automatic (“symmetric”) and Subversion now determines for the user which type of merge is necessary, simplifying the process and eliminating conflicts caused by users making the wrong selection. Symmetric merging handles changes simultaneously between branches rather than differentiating between the two merge forms – sync or reintegrate. In addition, Subversion 1.8 rejects attempts to merge between unrelated branches, decreasing the likelihood of user errors that often result in conflicts.

In addition, Subversion no longer requires the costly server set-up often cited in the past. Revision property packing and directory deltification, features of Subversion 1.8, reduce backup and restore times as well as the number of files stored, resulting in significant storage savings – up to more than 90% storage capacity savings. One company ran a test on its repository and storage decreased from 600+GB to just 17GB. Furthermore, a properly configured server allows for better performance on large projects due to extensive caching and other techniques (an area where SVN 1.8 also brings massive improvements).

We are a major sponsor of the Apache Software Foundation (ASF) and have made Subversion 1.8 release candidates (RedHat, SUSE and Debian versions) available for free download at

We will also make certified and fully tested Subversion 1.8 binaries available for download as soon as the ASF announces their general availability.

Following the pre-release of Subversion 1.8, we held a webinar, “Introducing Subversion 1.8,” hosted by leading technology industry journalist Adrian Bridgwater, and core Subversion developers Julian Foad and Philip Martin. They unveiled Subversion’s latest major release and provided an overview of its significant new features focusing on improvements to its merge capabilities.  Visit the replay page to view “Introducing Subversion 1.8“.

Topics include:

  • Automatic (“symmetric”) merge capability for simplifying the merge process and eliminating conflicts caused by users selecting the wrong type of merge.

  • Simultaneous change handling between branches rather than differentiating between sync and reintegration merge forms.

  • Rejection of merge attempts between unrelated branches to decrease the likelihood of user errors that often result in conflicts.

  • Decrease in server set-up costs.

  • Revision property packing and directory deltification to reduce backup and restore times as well as the number of files stored for up to more than 90% storage capacity savings.


“This free webinar is further proof of WANdisco’s continued dedication to the Apache Subversion community,” said David Richards, CEO of WANdisco. “Attendees gain great insight by hearing directly from committers who created the software and being able to ask them questions at the end of the presentation. That is a rare opportunity.”


uberSVN Update for ‘Chimney House’ Users

We’re pleased to announce an update to uberSVN ‘Chimney House’ that includes a new and improved manageAPPS page and LDAP enhancements.

uberSVN ‘Chimney House’ 3 features plenty of improvements, including:

  • Improvements to uberSVN APIs and internal development of uberSVN SDK (public release coming soon!)
  • New manageAPPS page allows you to see metadata attached to your APP license, such as expiry date, number of named users, and more.
  • Further improvements to the way uberSVN handles LDAP and LDAPS.
  • The latest Apache Subversion 1.7.5 binaries set to active by default.
  • A list of bug fixes, including some fixes and alignment of the uberSVN Access Control Team Leader and uberSVN Delegated Team Admin (where uberSVN Access Control is active)

You may have already heard, but with the latest release of Chimney House, we’re splitting uberSVN’s release cycle into two distinct phases. At least a few weeks before an update is released to the entire uberSVN user base, we’ll be giving our Latest Release Channel users a sneak preview of upcoming features and functionality. These users will get to test new features and see how they fit into the ALM environment before the update becomes widely available. Interested? Check out our blog post announcing the Latest Release Channel for more info.

For the full list of bug fixes, new features and improvements in uberSVN ‘Chimney House’ Release 3, see our Release Notes.

Not yet using uberSVN? It can be downloaded for free from

Our Initial Thoughts

Where did February go?  As a resident of Massachusetts, parts of me (namely my back, shoulders and arms) are quite happy to see those weeks get crossed off the calendar.  Since my last posting here, I’ve not only been shoveling snow though.   Two separate Subversion Live events were held,first in San Jose and then two weeks later in London.

Both days were extremely fruitful for those of us from WANdisco that attended and I hope our attendees felt the same way.   The presentations in the various tracks were well attended, very professional and well received. A great source of information exchange was the roundtable sessions at the conclusion of each day. But what I always find most useful at events like this are the less formal conversations that occur prior to a session, or during a break, or over lunch.  Not surprisingly, inquiries about what our committers were looking to do with regard to enhancing Subversion merge support was the most frequent topic raised and the subsequent discussions and feedback we received was extremely valuable.

So with that information in hand, here’s what is being initially targeted:

  • Better handling of renames across merges
  • partially automated merges
  • Faster merges – improving performance of merge operations
  • Enhancements to importing to handle 3rd Party / Vendor source code

This is really just an initial list. The team is actually still quite busy at the moment working on the final aspects of Subversion 1.7.  But we are also still in active discussions about other use cases that have been raised and additional ideas may still be formulated and as those become more concrete, I’ll be quite happy to write about them here.

By the way, if you are still interested in attending Subversion Live, February’s weather pattern here in the Northeastern U.S. has accommodated you! Our Boston event scheduled for early February has been rescheduled for Tuesday, March 22. The same great agenda awaits but thankfully, the snow may not. I hope to see many of you in Boston and I’d welcome the chance to listen and learn from your experience.

– Rob

P.S.  I really did have to shovel a lot, including my roof!

A Wealth of Information

As I stated in my last entry, there are many rich sources of feedback captured from the Subversion community.  The most obvious source of data is the Subversion project’s issue tracker, which contains a lot of the issues that will drive future updates to Subversion.  But there is a wealth of other data to look at to find what may be hindering users of Subversion.   The Subversion mailing lists and community sites can be particularly helpful in spotting recent trends, frustrations or simply common questions that continue to arise.

But one immediate source to focus our attention on is a couple of feedback sessions that were held about a year ago and hosted by Hyrum Wright and C. Michael Pilato.  The most interesting comment came from Mike’s session:

* Improved branching and merging: Everybody loves merge tracking
  and tree conflicts.  That is, when they don't hate it.
  Subversion should be smarter, and *must* learn to gracefully deal
  with renames.

Needless to say, we are already targeting better ways to handle merges across renames but it never hurts to see that reiterated in the summary of a feedback session such as this.

With respect to merge tracking in general, there are already a well known set of requirements that were captured by the project prior to the improvements delivered in Subversion 1.5.  These are a good reference point to start from but they do reflect the state of the project essentially at 1.4 and we want to update the use cases and capture relevant new use cases.  And while merging is a primary target for continued improvement, it is not the sole focus of this scoping exercise.  By no means are we limiting our attention to just this one topic.

We will continue to review both customer feedback and also the more specific issues captured in the issue tracker as we compile a prioritized list of targeted improvements. Beyond all of this data, I’m working with developers that have been involved in the project for years and their experience and instincts may prove to be as valuable source of ideas as any.

More to come…

– Rob

The Work Begins

As many reading this may already be aware, WANdisco has announced our intentions to focus our efforts towards continued enhancements to Subversion’s support for branching and merging. As our committers can see the end in sight for their efforts on Subversion 1.7, we are beginning to decide where next to direct our energy.

There already exists a wealth of data in various locations inside and outside of the project that capture good ideas for continuing the improvements to merging in Subversion. What we are not looking to do is to come up with many more ideas on the topic. Instead, we want to review and prioritize what is already out there and  then decide where we can best apply our resources for biggest benefit to the Subversion community.

The work we are doing now is to review the issues, capture the use cases and then do the hard work of applying our committers expertise towards resolving these issues. This process will very likely be incremental. Delivering new enhancements will provide benefits for many of us but it also allows for a new feedback loop to show us what remains to be done and where the most benefits lie for the next set of efforts. What we won’t be doing is trying to do something grandiose that could result in elongated development cycles and delayed delivery of some solutions.

The work has really just begun. We’re now rolling up our sleeves and diving into the work.  It should be fun!

I am planning to blog regularly about the process – your feedback is always welcome.

– Rob


About rbudas

Rob Budas has over 25 years of software industry experience, with the last 15 years focused on the Software Configuration Management sector. Prior to joining WANdisco, Rob had worked at IBM Rational for 8 years where he was a Sr. Product Manager for Rational ClearCase. He has held various development, technical sales and product management roles throughout his career. Rob holds a Bachelor of Science in Computer and Communication Science from the University of Michigan.