Mercurial > hg > hg-git
changeset 11:f2826f7b1ae5
sped up large imports significantly by caching parsed trees and sha_to_hexes
| author | Scott Chacon <schacon@gmail.com> |
|---|---|
| date | Sun, 26 Apr 2009 11:44:28 -0700 |
| parents | 66860f141788 |
| children | 227b11d75844 |
| files | TODO.txt __init__.py dulwich/objects.py dulwich/repo.py git_handler.py |
| diffstat | 5 files changed, 18 insertions(+), 13 deletions(-) [+] |
line wrap: on
line diff
--- a/TODO.txt +++ b/TODO.txt @@ -5,6 +5,7 @@ * update/add bookmarks * checkout the tip * limit to HEAD branch? (gh-pages makes weird import) +* some sort of remote management * tag conversion
--- a/__init__.py +++ b/__init__.py @@ -14,16 +14,6 @@ ''' -# -# Stage One - use Git commands to do the import / pushes, all in one big uggo file -# -# Stage Two - implement the Git packfile generation and server communication -# in native Python, so we don't need Git locally and don't need -# to keep all the git repo data around. We should just need a SHA -# mapping - since everything is append only in both systems it should -# be pretty simple to do. -# - # just importing every damn thing because i don't know python that well # and I have no idea what I actually need from mercurial import util, repair, merge, cmdutil, commands, hg, url @@ -55,7 +45,7 @@ git.fetch(git_url) # checkout the tip - # hg.update(ui, dest_repo) + hg.update(dest_repo, None) def gpush(ui, repo): dest_repo.ui.status(_("pushing to git url\n"))
--- a/dulwich/objects.py +++ b/dulwich/objects.py @@ -50,10 +50,17 @@ return dcomped +# SC hacked this to keep a global dict of already hexed shas because the +# import script calls this a bajillion times. Will try to cache other areas +# so this isn't called as much in the first place. +already_hexed_shas = {} def sha_to_hex(sha): """Takes a string and returns the hex of the sha within""" + if sha in already_hexed_shas: + return already_hexed_shas[sha] hexsha = "".join(["%02x" % ord(c) for c in sha]) assert len(hexsha) == 40, "Incorrect length of sha1 string: %d" % hexsha + already_hexed_shas[sha] = hexsha return hexsha
--- a/dulwich/repo.py +++ b/dulwich/repo.py @@ -296,8 +296,14 @@ def commit(self, sha): return self._get_object(sha, Commit) + # we call this a lot on import, so we're caching it a bit + already_parsed_trees = {} def tree(self, sha): - return self._get_object(sha, Tree) + if sha in self.already_parsed_trees: + return self.already_parsed_trees[sha] + tree = self._get_object(sha, Tree) + self.already_parsed_trees[sha] = tree + return tree def tag(self, sha): return self._get_object(sha, Tag)