Patchwork [7,of,7,V3] revset: introduce an API that avoids `formatspec` input serialization

login
register
mail settings
Submitter Boris Feld
Date Jan. 14, 2019, 2:27 p.m.
Message ID <466ba919947b7ced476c.1547476043@localhost.localdomain>
Download mbox | patch
Permalink /patch/37735/
State Accepted
Headers show

Comments

Boris Feld - Jan. 14, 2019, 2:27 p.m.
# HG changeset patch
# User Boris Feld <boris.feld@octobus.net>
# Date 1546605681 -3600
#      Fri Jan 04 13:41:21 2019 +0100
# Node ID 466ba919947b7ced476c3669ac444f0b74ade8f0
# Parent  1def212ed730ce024963fd41f2d341f584521bbf
# EXP-Topic revs-efficiency
# Available At https://bitbucket.org/octobus/mercurial-devel/
#              hg pull https://bitbucket.org/octobus/mercurial-devel/ -r 466ba919947b
revset: introduce an API that avoids `formatspec` input serialization

Instead of having the data fully serialized, the input can be directly inserted
in the tree at a later stage.

Just using it for simple "%ld" case provide a significant boost. For example
here are the impact on a sample discovery run between two pypy repositories
with arbitrary differences (using hg perfdiscovery).

$ hg perfdiscovery
before: ! wall 0.700435 comb 0.710000 user 0.700000 sys 0.010000 (median of 15)
after:  ! wall 0.501305 comb 0.510000 user 0.490000 sys 0.020000 (median of 20)
Yuya Nishihara - Jan. 15, 2019, 1:01 p.m.
On Mon, 14 Jan 2019 15:27:23 +0100, Boris Feld wrote:
> # HG changeset patch
> # User Boris Feld <boris.feld@octobus.net>
> # Date 1546605681 -3600
> #      Fri Jan 04 13:41:21 2019 +0100
> # Node ID 466ba919947b7ced476c3669ac444f0b74ade8f0
> # Parent  1def212ed730ce024963fd41f2d341f584521bbf
> # EXP-Topic revs-efficiency
> # Available At https://bitbucket.org/octobus/mercurial-devel/
> #              hg pull https://bitbucket.org/octobus/mercurial-devel/ -r 466ba919947b
> revset: introduce an API that avoids `formatspec` input serialization

> +    elif op in ('string', 'symbol', 'smartset'):

Update the tuple in foldconcat() just for clarity. These tuples looked
similar.

Patch

diff --git a/mercurial/localrepo.py b/mercurial/localrepo.py
--- a/mercurial/localrepo.py
+++ b/mercurial/localrepo.py
@@ -1362,9 +1362,8 @@  class localrepository(object):
         Returns a revset.abstractsmartset, which is a list-like interface
         that contains integer revisions.
         '''
-        expr = revsetlang.formatspec(expr, *args)
-        m = revset.match(None, expr)
-        return m(self)
+        tree = revsetlang.spectree(expr, *args)
+        return revset.makematcher(tree)(self)
 
     def set(self, expr, *args):
         '''Find revisions matching a revset and emit changectx instances.
diff --git a/mercurial/revset.py b/mercurial/revset.py
--- a/mercurial/revset.py
+++ b/mercurial/revset.py
@@ -125,6 +125,13 @@  def stringset(repo, subset, x, order):
         return baseset([x])
     return baseset()
 
+def rawsmartset(repo, subset, x, order):
+    """argument is already a smartset, use that directly"""
+    if order == followorder:
+        return subset & x
+    else:
+        return x & subset
+
 def rangeset(repo, subset, x, y, order):
     m = getset(repo, fullreposet(repo), x)
     n = getset(repo, fullreposet(repo), y)
@@ -2216,6 +2223,7 @@  methods = {
     "ancestor": ancestorspec,
     "parent": parentspec,
     "parentpost": parentpost,
+    "smartset": rawsmartset,
 }
 
 subscriptrelations = {
diff --git a/mercurial/revsetlang.py b/mercurial/revsetlang.py
--- a/mercurial/revsetlang.py
+++ b/mercurial/revsetlang.py
@@ -333,7 +333,7 @@  def _analyze(x):
     elif op == 'negate':
         s = getstring(x[1], _("can't negate that"))
         return _analyze(('string', '-' + s))
-    elif op in ('string', 'symbol'):
+    elif op in ('string', 'symbol', 'smartset'):
         return x
     elif op == 'rangeall':
         return (op, None)
@@ -373,7 +373,7 @@  def _optimize(x):
         return 0, x
 
     op = x[0]
-    if op in ('string', 'symbol'):
+    if op in ('string', 'symbol', 'smartset'):
         return 0.5, x # single revisions are small
     elif op == 'and':
         wa, ta = _optimize(x[1])
@@ -691,6 +691,29 @@  def formatspec(expr, *args):
             raise error.ProgrammingError("unknown revspec item type: %r" % t)
     return b''.join(ret)
 
+def spectree(expr, *args):
+    """similar to formatspec but return a parsed and optimized tree"""
+    parsed = _parseargs(expr, args)
+    ret = []
+    inputs = []
+    for t, arg in parsed:
+        if t is None:
+            ret.append(arg)
+        elif t == 'baseset':
+            newtree = ('smartset', smartset.baseset(arg))
+            inputs.append(newtree)
+            ret.append("$")
+        else:
+            raise error.ProgrammingError("unknown revspec item type: %r" % t)
+    expr = b''.join(ret)
+    tree = _parsewith(expr, syminitletters=_aliassyminitletters)
+    tree = parser.buildtree(tree, ('symbol', '$'), *inputs)
+    tree = foldconcat(tree)
+    tree = analyze(tree)
+    tree = optimize(tree)
+    return tree
+
+
 def _parseargs(expr, args):
     """parse the expression and replace all inexpensive args