revlog: make compressed size comparisons consistent
authorGregory Szorc <gregory.szorc@gmail.com>
Mon, 02 Jan 2017 11:50:17 -0800
changeset 30792 4215dc1b708b
parent 30791 ada160a8cfd8
child 30793 b6f455a6e4d6
revlog: make compressed size comparisons consistent revlog.compress() compares the compressed size to the input size and throws away the compressed data if it is larger than the input. This is the correct thing to do, as storing compressed data that is larger than the input takes up more storage space and makes reading slower. However, the comparison was implemented inconsistently. For the streaming compression mode, we threw away the result if it was greater than or equal to the input size. But for the one-shot compression, we threw away the compression only if it was greater than the input size! This patch changes the comparison for the simple case so it is consistent with the streaming case. As a few tests demonstrate, this adds 1 byte to some revlog entries. This is because of an added 'u' header on the chunk. It seems somewhat wrong to increase the revlog size here. However, IMO the cost of 1 byte in storage is insignificant compared to the performance gains of avoiding decompression. This patch should invite questions around the heuristic for throwing away compressed data. For example, I'd argue we should be more liberal about rejecting compressed data, additionally doing so where the number of bytes saved fails to reach a threshold. But we can have this discussion another time.
mercurial/revlog.py
tests/test-bundle.t
tests/test-commit-amend.t
--- a/mercurial/revlog.py	Sat Jan 07 20:43:49 2017 -0800
+++ b/mercurial/revlog.py	Mon Jan 02 11:50:17 2017 -0800
@@ -1503,7 +1503,7 @@
                 bin = "".join(p)
         else:
             bin = _compress(text)
-        if bin is None or len(bin) > l:
+        if bin is None or len(bin) >= l:
             if text[0] == '\0':
                 return ("", text)
             return ('u', text)
--- a/tests/test-bundle.t	Sat Jan 07 20:43:49 2017 -0800
+++ b/tests/test-bundle.t	Mon Jan 02 11:50:17 2017 -0800
@@ -268,13 +268,13 @@
 packed1 is produced properly
 
   $ hg -R test debugcreatestreamclonebundle packed.hg
-  writing 2663 bytes for 6 files
+  writing 2664 bytes for 6 files
   bundle requirements: generaldelta, revlogv1
 
   $ f -B 64 --size --sha1 --hexdump packed.hg
-  packed.hg: size=2826, sha1=e139f97692a142b19cdcff64a69697d5307ce6d4
+  packed.hg: size=2827, sha1=9d14cb90c66a21462d915ab33656f38b9deed686
   0000: 48 47 53 31 55 4e 00 00 00 00 00 00 00 06 00 00 |HGS1UN..........|
-  0010: 00 00 00 00 0a 67 00 16 67 65 6e 65 72 61 6c 64 |.....g..generald|
+  0010: 00 00 00 00 0a 68 00 16 67 65 6e 65 72 61 6c 64 |.....h..generald|
   0020: 65 6c 74 61 2c 72 65 76 6c 6f 67 76 31 00 64 61 |elta,revlogv1.da|
   0030: 74 61 2f 61 64 69 66 66 65 72 65 6e 74 66 69 6c |ta/adifferentfil|
 
--- a/tests/test-commit-amend.t	Sat Jan 07 20:43:49 2017 -0800
+++ b/tests/test-commit-amend.t	Mon Jan 02 11:50:17 2017 -0800
@@ -1147,7 +1147,7 @@
   R olddirname/newfile.py
   $ hg debugindex newdirname/newfile.py
      rev    offset  length  delta linkrev nodeid       p1           p2
-       0         0      88     -1       3 34a4d536c0c0 000000000000 000000000000
+       0         0      89     -1       3 34a4d536c0c0 000000000000 000000000000
 
   $ echo a >> newdirname/commonfile.py
   $ hg ci --amend -m bug
@@ -1155,7 +1155,7 @@
   newdirname/newfile.py renamed from olddirname/newfile.py:690b295714aed510803d3020da9c70fca8336def (glob)
   $ hg debugindex newdirname/newfile.py
      rev    offset  length  delta linkrev nodeid       p1           p2
-       0         0      88     -1       3 34a4d536c0c0 000000000000 000000000000
+       0         0      89     -1       3 34a4d536c0c0 000000000000 000000000000
 
 #if execbit