diff mercurial/tags.py @ 42237:9f45d3d526f9

hgtagsfnodescache: inherit fnode from parent when possible

If a changeset does not update the content of `.hgtags`, it means it will use
the same file-node (for `.hgtags`) as its parents. In this case we can
directly reuse the parent's file-node.

We use this property when updating the `hgtagsfnodescache` taking a faster path
if we already have a cached value for the parents of the node we are looking
at.

Doing so provides a large performance boost when looking at a lot of fnodes,
especially on repository with very large manifest:

timing for `tagsmod.fnoderevs(ui, repo, repo.changelog.revs())`

mercurial: (41907 revisions, 1923 files)

    before: 6.9 seconds
    after:  2.7 seconds (-54%)

pypy: (96266 revisions, 5198 files)

    before: 80 seconds
    after:  20 seconds (-75%)

mozilla-central: (463411 revisions, 272080 files)

    before: 7166.4 seconds
    after:    47.8 seconds (-99%, x150 speedup)

On a copy of mozilla-try with about 35K heads ans 1.7M changesets, this moves
the computation from many hours to a couple of minutes, making it more
interesting to do a full warm up of this cache before computing tags (from a
cold cache).

There seems to be other performance low hanging fruits, like avoiding the use of
changectx or a more revision centric logic. However, the new code is fast enough
for my needs right now.

--- a/mercurial/tags.py	Mon Mar 11 01:09:38 2019 +0100
+++ b/mercurial/tags.py	Mon Mar 11 01:10:20 2019 +0100
@@ -18,6 +18,7 @@
     bin,
     hex,
     nullid,
+    nullrev,
     short,
 )
 from .i18n import _
@@ -718,12 +719,33 @@
         if not computemissing:
             return None
 
-        # Populate missing entry.
-        try:
-            fnode = ctx.filenode('.hgtags')
-        except error.LookupError:
-            # No .hgtags file on this revision.
-            fnode = nullid
+        fnode = None
+        cl = self._repo.changelog
+        p1rev, p2rev = cl._uncheckedparentrevs(rev)
+        p1node = cl.node(p1rev)
+        p1fnode = self.getfnode(p1node, computemissing=False)
+        if p2rev != nullrev:
+            # There is some no-merge changeset where p1 is null and p2 is set
+            # Processing them as merge is just slower, but still gives a good
+            # result.
+            p2node = cl.node(p1rev)
+            p2fnode = self.getfnode(p2node, computemissing=False)
+            if p1fnode != p2fnode:
+                # we cannot rely on readfast because we don't know against what
+                # parent the readfast delta is computed
+                p1fnode = None
+        if p1fnode is not None:
+            mctx = ctx.manifestctx()
+            fnode = mctx.readfast().get('.hgtags')
+            if fnode is None:
+                fnode = p1fnode
+        if fnode is None:
+            # Populate missing entry.
+            try:
+                fnode = ctx.filenode('.hgtags')
+            except error.LookupError:
+                # No .hgtags file on this revision.
+                fnode = nullid
 
         self._writeentry(offset, properprefix, fnode)
         return fnode