largefiles: only cache largefiles in new heads

This fixes a serious performance regression in largefiles introduced in
Mercurial 2.1.  Caching new largefiles on pull is necessary, because
otherwise largefiles will be missing (and unable to be downloaded) when
the user tries to merge or rebase a new head with an old one.  But this
is an expensive operation and should only be done for heads that are new
from the pull, rather than on all heads in the repository.

hgext/largefiles/lfutil.py

@@ -449,3 +449,11 b' def mkstemp(repo, prefix):'
 class storeprotonotcapable(Exception):
     def __init__(self, storetypes):
         self.storetypes = storetypes
+
+def getcurrentheads(repo):
+    branches = repo.branchmap()
+    heads = []
+    for branch in branches:
+        newheads = repo.branchheads(branch)
+        heads = heads + newheads
+    return heads

hgext/largefiles/overrides.py

@@ -660,6 +660,7 b' def override_pull(orig, ui, repo, source'
         repo.lfpullsource = source
         if not source:
             source = 'default'
+        oldheads = lfutil.getcurrentheads(repo)
         result = orig(ui, repo, source, **opts)
         # If we do not have the new largefiles for any new heads we pulled, we
         # will run into a problem later if we try to merge or rebase with one of
@@ -667,10 +668,9 b' def override_pull(orig, ui, repo, source'
         # cache.
         ui.status(_("caching new largefiles\n"))
         numcached = 0
-        branches = repo.branchmap()
-        for branch in branches:
-            heads = repo.branchheads(branch)
-            for head in heads:
+        heads = lfutil.getcurrentheads(repo)
+        newheads = set(heads).difference(set(oldheads))
+        for head in newheads:
                 (cached, missing) = lfcommands.cachelfiles(ui, repo, head)
                 numcached += len(cached)
         ui.status(_("%d largefiles cached\n" % numcached))