treemanifests: remove _loadalllazy in _diff()

The benchmarks below use a similar setup as in ee7ee0c516ca and my other recent
commits.  Yes, in some cases this runs in literally 5% of the time it
previously took.

before = a6f8ab53

    diff --git:
    repo  | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
    ------+---+---+------------------------+-----------------------+------------
    m-u   |   |   | 1.301 s +-  0.004 s    | 1.309 s +-  0.012 s   | 100.6%
    m-u   |   | x | 1.303 s +-  0.009 s    | 1.302 s +-  0.006 s   |  99.9%
    m-u   | x |   | 1.308 s +-  0.006 s    | 1.309 s +-  0.007 s   | 100.1%
    m-u   | x | x | 85.7 ms +-   0.6 ms    | 86.0 ms +-   0.3 ms   | 100.4%
    l-d-r |   |   | 197.5 ms +-   0.7 ms   | 197.8 ms +-   2.2 ms  | 100.2%
    l-d-r |   | x | 199.4 ms +-   0.6 ms   | 199.3 ms +-   0.9 ms  |  99.9%
    l-d-r | x |   | 86.1 ms +-   0.5 ms    | 85.8 ms +-   0.9 ms   |  99.7%
    l-d-r | x | x | 64.4 ms +-   0.4 ms    | 64.4 ms +-   0.3 ms   | 100.0%

    diff -c . --git:
    repo  | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
    ------+---+---+------------------------+-----------------------+------------
    m-u   |   |   | 236.7 ms +-   1.1 ms   | 236.5 ms +-   1.3 ms  |  99.9%
    m-u   |   | x | 158.7 ms +-   1.0 ms   | 128.0 ms +-   1.0 ms  |  80.7% <--
    m-u   | x |   | 239.7 ms +-   1.8 ms   | 238.1 ms +-   1.5 ms  |  99.3%
    m-u   | x | x | 132.4 ms +-   0.9 ms   | 132.3 ms +-   0.6 ms  |  99.9%
    l-d-r |   |   | 81.8 ms +-   0.4 ms    | 81.8 ms +-   0.3 ms   | 100.0%
    l-d-r |   | x | 3.894 s +-  0.017 s    | 193.6 ms +-   0.4 ms  |   5.0% <--
    l-d-r | x |   | 106.9 ms +-   0.4 ms   | 106.6 ms +-   0.3 ms  |  99.7%
    l-d-r | x | x | 182.7 ms +-   0.8 ms   | 183.3 ms +-   0.9 ms  | 100.3%

    rebase -r . --keep -d .^^:
    repo  | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
    ------+---+---+------------------------+-----------------------+------------
    m-u   |   |   | 5.615 s +-  0.109 s    | 5.562 s +-  0.015 s   |  99.1%
    m-u   |   | x | 5.701 s +-  0.027 s    | 5.715 s +-  0.023 s   | 100.2%
    m-u   | x |   | 5.572 s +-  0.128 s    | 5.613 s +-  0.182 s   | 100.7%
    m-u   | x | x | 633.3 ms +-  28.7 ms   | 636.2 ms +-  13.8 ms  | 100.5%
    l-d-r |   |   | 666.4 ms +-  17.0 ms   | 658.5 ms +-   9.3 ms  |  98.8%
    l-d-r |   | x | 6.520 s +-  0.070 s    | 6.505 s +-  0.026 s   |  99.8%
    l-d-r | x |   | 279.0 ms +-  13.0 ms   | 276.5 ms +-   4.7 ms  |  99.1%
    l-d-r | x | x | 1.636 s +-  0.058 s    | 1.657 s +-  0.014 s   | 101.3%

    status --change . --copies:
    repo  | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
    ------+---+---+------------------------+-----------------------+------------
    m-u   |   |   | 218.6 ms +-   1.4 ms   | 217.9 ms +-   1.5 ms  |  99.7%
    m-u   |   | x | 138.5 ms +-   0.4 ms   | 108.4 ms +-   0.2 ms  |  78.3% <--
    m-u   | x |   | 220.1 ms +-   1.3 ms   | 219.7 ms +-   1.5 ms  |  99.8%
    m-u   | x | x | 113.2 ms +-   0.4 ms   | 112.4 ms +-   0.8 ms  |  99.3%
    l-d-r |   |   | 80.2 ms +-   0.3 ms    | 80.6 ms +-   0.6 ms   | 100.5%
    l-d-r |   | x | 3.899 s +-  0.020 s    | 194.8 ms +-   4.0 ms  |   5.0% <--
    l-d-r | x |   | 83.4 ms +-   0.8 ms    | 83.2 ms +-   0.2 ms   |  99.8%
    l-d-r | x | x | 732.2 ms +-   4.3 ms   | 194.9 ms +-   1.0 ms  |  26.6% <--

    status --copies:
    repo  | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
    ------+---+---+------------------------+-----------------------+------------
    m-u   |   |   | 1.917 s +-  0.005 s    | 1.914 s +-  0.004 s   |  99.8%
    m-u   |   | x | 1.909 s +-  0.012 s    | 1.934 s +-  0.004 s   | 101.3%
    m-u   | x |   | 1.915 s +-  0.005 s    | 1.904 s +-  0.004 s   |  99.4%
    m-u   | x | x | 94.8 ms +-   0.3 ms    | 94.7 ms +-   0.2 ms   |  99.9%
    l-d-r |   |   | 593.9 ms +-   1.2 ms   | 594.6 ms +-   9.4 ms  | 100.1%
    l-d-r |   | x | 595.2 ms +-   3.8 ms   | 597.2 ms +-   2.6 ms  | 100.3%
    l-d-r | x |   | 182.5 ms +-   1.6 ms   | 182.1 ms +-   0.6 ms  |  99.8%
    l-d-r | x | x | 149.6 ms +-   0.9 ms   | 149.1 ms +-   0.8 ms  |  99.7%

    update $rev^; ~/src/hg/hg{hg}/hg update $rev:
    repo  | N | T | before (mean +- stdev) | after (mean +- stdev) | % of before
    ------+---+---+------------------------+-----------------------+------------
    m-u   |   |   | 3.121 s +-  0.007 s    | 3.129 s +-  0.012 s   | 100.3%
    m-u   |   | x | 2.972 s +-  0.011 s    | 2.981 s +-  0.012 s   | 100.3%
    m-u   | x |   | 3.144 s +-  0.014 s    | 3.141 s +-  0.011 s   |  99.9%
    m-u   | x | x | 312.2 ms +-   2.4 ms   | 312.3 ms +-   2.1 ms  | 100.0%
    l-d-r |   |   | 444.4 ms +-   4.3 ms   | 446.9 ms +-   5.3 ms  | 100.6%
    l-d-r |   | x | 9.159 s +-  0.069 s    | 9.182 s +-  0.040 s   | 100.3%
    l-d-r | x |   | 254.6 ms +-   1.6 ms   | 255.2 ms +-   1.6 ms  | 100.2%
    l-d-r | x | x | 1.525 s +-  0.007 s    | 1.577 s +-  0.007 s   | 103.4% <--?

Differential Revision: https://phab.mercurial-scm.org/D4845

mercurial/manifest.py

@@ -1114,9 +1114,19 b' class treemanifest(object):'
                 return
             t1._load()
             t2._load()
-            # OPT: do we need to load everything?
-            t1._loadalllazy()
-            t2._loadalllazy()
+            toloadlazy = []
+            for d, v1 in t1._lazydirs.iteritems():
+                v2 = t2._lazydirs.get(d)
+                if not v2 or v2[1] != v1[1]:
+                    toloadlazy.append(d)
+            for d, v1 in t2._lazydirs.iteritems():
+                if d not in t1._lazydirs:
+                    toloadlazy.append(d)
+
+            for d in toloadlazy:
+                t1._loadlazy(d)
+                t2._loadlazy(d)
+
             for d, m1 in t1._dirs.iteritems():
                 m2 = t2._dirs.get(d, emptytree)
                 _diff(m1, m2)