Is version control (ie. Subversion) applicable in document tracking? [closed]

I've worked with Word documents in SVN. With TortoiseSVN, you can easily diff Word documents (between working copy and repository, or between two repository revisions). It's really slick and definitely recommended.

The other thing to do if you're using Word documents in SVN is to add the svn:needs-lock property to the Word documents. This will prevent two people from trying to edit the same document at the same time, since unfortunately there's no good way to merge Word documents.

With the above two things, handling revision controlled Word documents is at least tolerable. It certainly beats the alternative of using a shared folder and track-changes.


What on Earth are you all Word-is-binary-so-no-diff people talking about? TortoiseSVN, for example, integrates right out of the box with Word and enables you to use Word's built-in diff and merge functionality. It works just fine.

I have worked on projects that store documents in version control. It has worked out pretty well, although if people are unfamiliar with version control, they are probably going to have conceptual difficulties with things like "working copy" and "merge" and "conflict". Don't overestimate the users' capabilities when you plan your document management system.

I believe there exist big and powerful commercial solutions for all of this, as well. I'm sure if you have enough kilodollars, you can get something that fits your needs perfectly. Document management systems are a big business for big enterprise.


I guess one thing that nobody seems to have asked is if you have a legal requirement to store history of changes to the doc's?

Whether you do or don't is going to have an impact on what solutions you can consider.

Also a notification mechanism for out of date copies is also a bundle of fun. If engineer A has a copy of a document and engineer B then edits it and commits the changes you want engineer A to be notified that his copy is out of date.

Document control can become a real can of worms quite easily.

Maybe keep the doc's under CVS or SVN and set it up so that emails are generated to whoever has checked out a copy when updates for the same doc. are checked in to the repository?

Edit: I forgot to add don't forget to use the binary switch, e.g. -kb for CVS, when adding the new doc. Otherwise, you will get any sequences of data that happen to match the ascii for keyword strings having the relevant config management data appended thereby corrupting your doc. data.


Thinking out of the box, would migrating to a Wiki be out of the question?

Since you consider it feasible to force your users into Subversion (or something similar), a larger change seem acceptable.

Another migration target could be to use some kind of structured XML document format (DocBook comes to mind). This would enable you to indeed use diffs and source control, while getting all sorts of document formats for free.