Comment 3 for bug 145155

At the moment we're treating tree metadata as actual data and that means we'll end up with two copies of the ignores in subversion (one in .bzrignore, one in their file properties) to import from. Importing from svn after that becomes hard because it means you have to do merging and potential conflict resolution.

Having .bzrignore as data also means we have to be able to regenerate the .bzrignore file from the svn properties with the exact same contents each time and with ordering preserved. A non-workingtree-file could not have any restrictions wrt ordering.

Finding the ignores is also O(tree) right now since the ignores file would have to be generated on the fly.

There was another issue, but I seem to've forgotten that one.