hglib vs dulwich finding blobs
ActivePublic
Actions

Authored by fiendish on Apr 4 2017, 4:18 PM.

Tags

None

Subscribers

None

	# hgrepo and gitrepo are identical except hg and git.
	# all times are on a fast SSD

	# this takes about 20 seconds
	hgblobs = {}
	for li in hgrepo.log():
	c = hgrepo[li]
	fs = c.added()+c.modified()
	for f in fs:
	data = hgrepo.cat([os.path.join(hgrepo.root(), f)], c.rev())
	hgblobs[data] = (f, c.rev())


	# this (better!) takes about 14 seconds
	hgblobs = {}
	for f in hgrepo.manifest(all=True):
	fpath = os.path.join(hgrepo.root(), f)
	for li in hgrepo.log(files=[fpath]):
	data = hgrepo.cat([fpath], int(li[0]))
	hgblobs[data] = (f, li[0])


	# this takes about 3 seconds
	gitblobs = [v for k in gitrepo.object_store.packs for v in k.iterobjects() if v.type_name == b'blob']


	# There are 1686 revs in the repos. 2374 blobs.

fiendish edited the content of this paste. (Show Details)

hglib vs dulwich finding blobsActivePublicActions