How to save index with "cache tree" extension?

Hello,

I have a project with approx 2000 files stored on NFS partition. When I initialize repository with pygit2 all subsequent commits took about 5sec. But after I commit anything (even empty tree) via git subsequent commits via pygit2 took only 0.5sec. This is because commit via git writes cache tree (https://git-scm.com/docs/index-format#_cache_tree) extension to .git/index file.

Finally I found that reading tree back to the index builds internal "cache tree". And now I wondering if this is a correct way to buid cache tree on initial commit?

repo = pygit2.init_repository(path, bare=False, flags=pygit2.GIT_REPOSITORY_INIT_NO_REINIT)
index = repo.index
index.add_all()
tree_id = index.write_tree()
# NOTE: Adds "cache tree" extension to the index.
index.read_tree(tree_id)
index.write()
repo.create_commit("HEAD", SIGNATURE, SIGNATURE, "Inital commit", tree_id, [])

Comienza con las llamadas a pygit2.init_repository, index.add_all(), index.write_tree(), index.read_tree(), index.write() y repo.create_commit() mostradas en el issue; después, inspecciona cómo se gestionan las index extensions. Compara el índice resultante con uno producido por git en un repositorio con muchos archivos. Se considera terminado establecer si cache tree es compatible y documentar o corregir el comportamiento del commit inicial.