I'm looking for some information about the differences between traditional software CM and CM for web development if there is any. I just found out that CM is recommended for web development but I did't find out how it has to be performed nor if it is different from traditional SCM. What do you think are the differences?
-- query posted on CMCrossRoads General CM Forum

"Web development" can be everything from a simple intranet site to a high-profile e-commerce internet site. However, regardless, the response to whether CM for web is different than traditional CM is "No."

The most common reason why web and traditional CM are often looked at differently is because management, particularly project management, often confuses configuration management and content management and assumes that they are both the responsibility of and can be carried out by the SCM group and their traditional tools.

The fundamental differentiator between configuration and content management is that the latter places a stronger emphasis on understanding what the content is (its meaning, relevance, personalization, targets, relationships, etc.) rather than just managing it at the object level. The traditional SCM tools have big problems in dealing with content volume, probably the most the most underestimated driver in web development. Websites, intranets, B2B extranets, on-line catalogues, multimedia, etc., can produce thousands of pages daily. It is virtually impossible to manage the sheer volume of information and the rate of change without some form of automated content processes and tools, not those designed for SCM.

The other problem that the majority of my clients experience when attempting to combine the two disciplines is the streamlined delivery required. The immeasurable volume of information published daily on the web means that traditional SCM review lifecycles are an unaffordable luxury. Content management provides the key tools required to expedite and orchestrate those types of delivery processes.

Do not get me wrong, there is a need to create and capture content, a need to store it, retrieve it, manipulate it, deliver it, and ultimately enhance its value through update and reuse. This sounds very much like tradiitional SCM and SCM can be very helpful complimenting content management, not replacing it. An example that I see time and time again is that graphic content is developed using tools that have very proprietary outputs. When the graphic is to be placed on the web, it is often transposed into either a .jpeg or .gif. The original drawing should be versioned in a SCM tool (Very often isn't) while the transposed image becomes part of content where it can be deployed instantly to the targets that need it. At some later date, if the orginal needs to be modified, it can be checked-out in its native format without having to recreate it from scratch.

-- DonPeck^? - posted 02 Jun 2003 to CMCrossRoads General CM Forum

Configuration managment deals with both space and time. Space is the view of a configuration at any given time. Time is versioning, branching, and merging. You can think of it like parallel universes in Star Trek -- just like ours, but a little different.

The main think to think about in how you devide up your space is "What will be versioned?" That is, what collection of things will be combined together to have a versioned identity over time. Typically, dividing things along the type of source (e.g. static versus dynamic) does not necessarily fit the logical grouping for versioning. For example, logically you might have related static and dynamic web pages which should be treated as one thing for a single purpose.

Sometimes there are reasons to split out some types of files. Binaries may be handled differently than text. In ClearCase, text is stored efficiently as deltas and can be merged. Binaries are stored less effiently (for space) and the only "merging" is to pick one version or the other. For this reason, it might be desirable to revise and branch binaries less often than text files. For web space, I think you would be largely text based so binaries may not be a worry for you.

An interesting aspect of your posting is that you are not in control of the changes, the customer is. Most CM systems have a concept of checkout. As such, the system is actually recording deltas. If you customers are modifying the files, you really run the risk of not knowing what they started with. Resolving changes from multiple sources could be difficult.

An interesting article appears in this months Scientific American entitled Chain letters and Evolutionary Histories. In the article, the authors propose a clever scheme for measuring the distance between two variations. Basically, they compress A, compress B, and compress A concatenated to B and measure the relative size savings. The more A and B have in common, the more AB will compress. They use a utility called gencompress used on the human genome project.

So, if you have a way for your costomers to checkout, modify, and checkin that is the easiest to manage. But if not, something like the above might be of interest. And if nothing else, the article is entertaining!

-- JoeCaporaletti^?