# HG changeset patch
# User Jun Wu <quark@fb.com>
# Date 1464705851 -3600
# Node ID b6e2bc9625365c2dfc1487230369f9ed79a20924
# Parent  7bb2c6ca4d2497f6107cfc553a193cbb99fe37b5
maps: increase sqlite cache size automatically

Sqlite cache size is 2MB by default. Increasing the cache size can speed up
migration. In a large repo with millions of revisions, changing the cache
size to a big enough value reduces at least 30% of the migration time.

Give the general trends that sacrifices space for speed, this patch auto
tunes the cache size to be big enough to hold the entire database.

diff --git a/hgsubversion/maps.py b/hgsubversion/maps.py
--- a/hgsubversion/maps.py
+++ b/hgsubversion/maps.py
@@ -708,6 +708,15 @@ class SqliteRevMap(collections.MutableMa
         self._db = sqlite3.connect(self._dbpath)
         self._db.text_factory = bytes
 
+        # cache size affects random accessing (e.g. index building)
+        # performance greatly. default is 2MB (2000 KB), we want to have
+        # a big enough cache that can hold the entire map.
+        cachesize = 2000
+        for path, ratio in [(self._filepath, 1.7), (self._dbpath, 1)]:
+            if os.path.exists(path):
+                cachesize += os.stat(path).st_size * ratio // 1000
+        self._db.execute('PRAGMA cache_size=%d' % (-cachesize))
+
         # disable auto-commit. everything is inside a transaction
         self._db.isolation_level = 'DEFERRED'