Mercurial > hg
comparison mercurial/changegroup.py @ 30211:6b0741d6d234
changegroup: skip delta when the underlying revlog do not use them
Revlog can now be configured to store full snapshot only. This is used on the
changelog. However, the changegroup packing was still recomputing deltas to be
sent over the wire.
We now just reuse the full snapshot directly in this case, skipping delta
computation. This provides use with a large speed up(-30%):
# perfchangegroupchangelog on mercurial
! wall 2.010326 comb 2.020000 user 2.000000 sys 0.020000 (best of 5)
! wall 1.382039 comb 1.380000 user 1.370000 sys 0.010000 (best of 8)
# perfchangegroupchangelog on pypy
! wall 5.792589 comb 5.780000 user 5.780000 sys 0.000000 (best of 3)
! wall 3.911158 comb 3.920000 user 3.900000 sys 0.020000 (best of 3)
# perfchangegroupchangelog on mozilla central
! wall 20.683727 comb 20.680000 user 20.630000 sys 0.050000 (best of 3)
! wall 14.190204 comb 14.190000 user 14.150000 sys 0.040000 (best of 3)
Many tests have to be updated because of the change in bundle content. All
theses update have been verified. Because diffing changelog was not very
valuable, the resulting bundle have similar size (often a bit smaller):
# full bundle of mozilla central
with delta: 1142740533B
without delta: 1142173300B
So this is a win all over the board.
author | Pierre-Yves David <pierre-yves.david@ens-lyon.org> |
---|---|
date | Fri, 14 Oct 2016 01:31:11 +0200 |
parents | edb49a90723c |
children | 260af19891f2 |
comparison
equal
deleted
inserted
replaced
30210:5e4f16874a9f | 30211:6b0741d6d234 |
---|---|
816 # bundle.reorder=auto just like bundle.reorder=False). | 816 # bundle.reorder=auto just like bundle.reorder=False). |
817 self._reorder = False | 817 self._reorder = False |
818 | 818 |
819 def deltaparent(self, revlog, rev, p1, p2, prev): | 819 def deltaparent(self, revlog, rev, p1, p2, prev): |
820 dp = revlog.deltaparent(rev) | 820 dp = revlog.deltaparent(rev) |
821 # Avoid sending full revisions when delta parent is null. Pick | 821 if dp == nullrev and revlog.storedeltachains: |
822 # prev in that case. It's tempting to pick p1 in this case, as p1 | 822 # Avoid sending full revisions when delta parent is null. Pick prev |
823 # will be smaller in the common case. However, computing a delta | 823 # in that case. It's tempting to pick p1 in this case, as p1 will |
824 # against p1 may require resolving the raw text of p1, which could | 824 # be smaller in the common case. However, computing a delta against |
825 # be expensive. The revlog caches should have prev cached, meaning | 825 # p1 may require resolving the raw text of p1, which could be |
826 # less CPU for changegroup generation. There is likely room to add | 826 # expensive. The revlog caches should have prev cached, meaning |
827 # a flag and/or config option to control this behavior. | 827 # less CPU for changegroup generation. There is likely room to add |
828 # | 828 # a flag and/or config option to control this behavior. |
829 # Pick prev when we can't be sure remote has the base revision. | |
830 if dp == nullrev or (dp != p1 and dp != p2 and dp != prev): | |
831 return prev | 829 return prev |
832 return dp | 830 elif dp == nullrev: |
831 # revlog is configured to use full snapshot for a reason, | |
832 # stick to full snapshot. | |
833 return nullrev | |
834 elif dp not in (p1, p2, prev): | |
835 # Pick prev when we can't be sure remote has the base revision. | |
836 return prev | |
837 else: | |
838 return dp | |
833 | 839 |
834 def builddeltaheader(self, node, p1n, p2n, basenode, linknode, flags): | 840 def builddeltaheader(self, node, p1n, p2n, basenode, linknode, flags): |
835 # Do nothing with flags, it is implicitly 0 in cg1 and cg2 | 841 # Do nothing with flags, it is implicitly 0 in cg1 and cg2 |
836 return struct.pack(self.deltaheader, node, p1n, p2n, basenode, linknode) | 842 return struct.pack(self.deltaheader, node, p1n, p2n, basenode, linknode) |
837 | 843 |