Proper diffs with ODF files in Git or Mercurial | Hacker News Reader

Proper diffs with ODF files in Git or Mercurial

www-verimag.imag.fr

31 pointsbergie7 comments

Pretty cool.

Would would be even better would be if:

a) Git diff/merge gets ability to look into zip/archive files, treating them as sub-trees of files. diff should be not too tricky. Merge is more awkward, because you have to create a new zip/archive which is compatible with the format of the parents.

b) Git diff/merge gets ability to work well on XML files, and be XML-element-based, not line-based.

With those, full version control with branching and merging of ODF files, plus many other interesting file types, falls out almost "naturally".

git diff/merge on xml files would be really great. will think this over.

I'm not sure if odt2txt puts paragraphs on one line, or flows them. But either way, if you change a single word in a long paragraph you can get large-looking diffs. For version-controlled documents it's good to use a word-based diff rather than a line-based diff.

I use wdiff (although it's slightly kludgy) with documents and colorize the output: http://homepages.inf.ed.ac.uk/imurray2/compnotes/cwdiff/

'git diff' already supports --word-diff natively.

Great! Thanks. I now see that git also has a built-in --color-words option, and I've added that to my notes. (I don't have a git setup to quickly try whether that would still work in conjunction with an external diff tool like oodiff though.)

In the mean time I'm using my (trivial) cwdiff wrapper in place of diff in tools that don't have a built-in option. I think it could be called from the oodiff utility here, where it would work equally well with any versioning system.

It seems to flow them. At least that is what I see with git log -p.

The benefit with this approach is that it works with exactly the same commands that I use for source control.

I'm thinking of building a productivity tool in the legal space. The big inspiration I had a week ago is that I could use git or another version control system for features like git-diff and git-blame.

I'm now thinking about how most elegantly add markup features for things like highlighting text, adding labels, inline commenting, muti-media embeds, and text formatting both simple and complex. any suggestions? The naive thing to do is to include a <span> based markup format that complements what is already being done with font-size,color etc but im worried that it will become a mess. Both in terms of document size and git diff.

also, are there any examples of products that successfully leveraged version control systems to create large project management tools?

Reply on news.ycombinator.com