Cache the non-recursive file listing

This in turn means that commits on those files are also cached, which
are the more expensive operation.  (Because finding commits for files
involves walking and diffing the history.)

quit.go

func (b byTypeAndName) Swap(i, j int)      { b[i], b[j] = b[j], b[i] }
func (fc *FancyCommit) Files(recursive bool) ([]*FancyFile, error) {
	tree, err := fc.commit.Tree()
	if err != nil {
		return nil, err
	}
	if recursive || fc.files == nil {
		tree, err := fc.commit.Tree()
		if err != nil {
			return nil, err
		}
	recurse := 1
	if recursive {
		recurse = 0
	}
	files := make([]*FancyFile, 0)
	err = tree.Walk(func(s string, entry *git.TreeEntry) int {
		files = append(files, NewFancyFile(fc.commit.Owner(), entry))
		return recurse
	})
	if err != nil {
		return nil, err
	}
		recurse := 1
		if recursive {
			recurse = 0
		}
		files := make([]*FancyFile, 0)
		err = tree.Walk(func(s string, entry *git.TreeEntry) int {
			files = append(files, NewFancyFile(fc.commit.Owner(), entry))
			return recurse
		})
		if err != nil {
			return nil, err
		}
	sort.Sort(byTypeAndName(files))
		sort.Sort(byTypeAndName(files))
	return files, nil
		if recursive {
			return files, nil
		}
		fc.files = files
	}
	return fc.files, nil
}
var readmeNames = []string{"README.md", "readme.md", "README.rst", "readme.rst", "README.txt", "readme.txt", "README"}