From: Avery Pennarun <apenwarr@gmail.com>
Date: Tue, 2 Feb 2010 02:30:59 +0000 (-0500)
Subject: cmd-midx: correctly handle a tiny nonzero number of objects.
X-Git-Url: https://git.michaelhowe.org/gitweb/?a=commitdiff_plain;h=11f86ebed112fa8305271388c71248a51740b644;p=packages%2Fb%2Fbup.git

cmd-midx: correctly handle a tiny nonzero number of objects.

If all the sha1sums would have fit in a single page, the number of bits in
the table would be negative, with odd results.  Now we just refuse to create
the midx if there are too few objects *and* too few files, since it would be
useless anyway.

We're still willing to create a very small midx if it allows us to merge
several indexes into one, however small, since that will still speed up
searching.
---

diff --git a/cmd-midx.py b/cmd-midx.py
index 97d01de..c0f8324 100755
--- a/cmd-midx.py
+++ b/cmd-midx.py
@@ -29,15 +29,15 @@ def do_midx(outdir, outfilename, infilenames):
         inp.append(ix)
         total += len(ix)
 
-    if not total:
-        log('%s: no new .idx files: nothing to do.\n' % outdir)
+    log('Merging %d indexes (%d objects).\n' % (len(infilenames), total))
+    if total < 1024 and len(infilenames) < 3:
+        log('%s: not enough objects for a .midx to be useful.\n' % outdir)
         return
 
-    log('Merging %d indexes (%d objects).\n' % (len(infilenames), total))
-    pages = total/SHA_PER_PAGE
+    pages = int(total/SHA_PER_PAGE) or 1
     bits = int(math.ceil(math.log(pages, 2)))
     entries = 2**bits
-    log('table size: %d (%d bits)\n' % (entries*8, bits))
+    log('Table size: %d (%d bits)\n' % (entries*4, bits))
     
     table = [0]*entries