Stephen Hahn wrote: > * Steve Cogorno <Steven.Cogorno at Sun.COM> [2006-01-18 10:11]: >> I'd strongly caution against using an SCM-type system for the XML >> files. >> This would work OK if the files were always edited using a standard >> text editor, but it will not work if any sort of GUI XML editor is >> used >> because the line breaks will change each time the file is saved. >> This >> won't allow proper diffs and hence merging will be almost impossible. >> >> Also, the method by which the documents will be exported from the Sun >> content repositories will not guarantee any sort of consistency of >> line >> breaks. From release to release of the base Sun documentation, the >> line >> breaks will be completely different. > > > One possibility is, prior to integration, to pass documents > through a > consistent formatting filter, much like indent(1) for C. > Constructing > a filter that had a reasonable set of defaults to minimize line > break > drift doesn't seem profoundly difficult. That is to say, > assuming the > toolchain to be unchangeable might be unnecessarily constraining.
Given the hierarchical nature of the content, it's much more difficult than you might think. For XML files that represent data files, it is pretty easy to do. It is not at all simple to do with text-based content. Keep in mind that there are more than 300 elements in SolBook. At least 100 of them are white-space significant. Given that there are many thousands of permutations of usage, this is a very difficult problem. It's not intractable, but it would be very time consuming and resource intensive. -Steve
