Output is incorrect. Race?

[artagnon]

Hi,

I'm doing this on the git.git repository.

$ git log >/tmp/log
$ git log | e-sink.pl # save the buffer as /tmp/log.2
$ git diff /tmp/log /tmp/log.2
diff --git a/tmp/log b/tmp/log.2
index 52b4dbc..f96165c 100644
--- a/tmp/log
+++ b/tmp/log.2
@@ -1,3 +1,4 @@
+<<<<< start: 2013-03-26T15:11:03
 commit 0eab1c9f686b6d6a901dd7e2631034e5234d2c1c
 Author: Ramkumar Ramachandra <[email protected]>
 Date:   Fri Mar 22 15:53:27 2013 +0530
@@ -151866,6 +151867,212 @@ Author: Michael J Gruber <[email protected]>
 Date:   Fri May 14 20:26:51 2010 +0200

     t6018: make sure all tested symbolic names are different revs
+t is not
+       important to paginate this, so don’t.
+    
+    Signed-off-by: Nguyễn Thái Ngọc Duy <[email protected]>
+    Signed-off-by: Jonathan Nieder <[email protected]>
+    Acked-by: Jeff King <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit bce2c9ae9ff458b6090953ab9f639255f757a104
+Author: Jonathan Nieder <[email protected]>
+Date:   Sat Jun 26 14:25:37 2010 -0500
+
+    tests: local config file should be honored from subdirs of toplevel
+    
+    When git is passed the --paginate option, starting up a pager requires
+    deciding what pager to start, which requires access to the core.pager
+    configuration.  If --paginate is handled before searching for the
+    git dir, this configuration will be missed.
+    
+    In other words, with --paginate and only with --paginate, any
+    repository-local core.pager setting is being ignored [*].
+    
+    [*] unless the git directory is ./.git or GIT_DIR or GIT_CONFIG was
+    set explicitly.
+    
+    Add a test to demonstrate this counterintuitive behavior.  Noticed
+    while reading over a patch by Duy that fixes it.
+    
+    Cc: Nguyễn Thái Ngọc Duy <[email protected]>
+    Improved-by: Johannes Sixt <[email protected]>
+    Signed-off-by: Jonathan Nieder <[email protected]>
+    Acked-by: Jeff King <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 8f81449e885ab2b9bac09b5b835314d26f107b3f
+Author: Jonathan Nieder <[email protected]>
+Date:   Sat Jun 26 14:24:50 2010 -0500
+
+    t7006: test pager configuration for several git commands
+    
+    Test choice of pager at several stages of repository setup.  This
+    provides some (admittedly uninteresting) examples to keep in mind when
+    considering changes to the setup procedure.
+    
+    Improved-by: Johannes Sixt <[email protected]>
+    Signed-off-by: Jonathan Nieder <[email protected]>
+    Acked-by: Jeff King <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 3c7406d4b595c9e8c8dfcff2739bab9412add5a3
+Author: Jonathan Nieder <[email protected]>
+Date:   Sat Jun 26 14:23:02 2010 -0500
+
+    t7006 (pager): introduce helper for parameterized tests
+    
+    The current tests test pager configuration for ‘git log’, but other
+    commands use a different setup procedure and should therefore be
+    tested separately.  Add a helper to make this easier.
+    
+    This patch introduces the helper and changes some existing tests to
+    use it.  The only functional change should be the introduction of ‘git
+    log - ’ to a few test descriptions.
+    
+    Signed-off-by: Jonathan Nieder <[email protected]>
+    Acked-by: Jeff King <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 89fe121d5fd808391ee38b7f39b88cb3f912776f
+Author: Ramsay Jones <[email protected]>
+Date:   Mon Jun 21 19:52:29 2010 +0100
+
+    notes: Initialise variable to appease gcc
+    
+    gcc version 3.4.4 thinks that the 'cmp' variable could be used
+    while uninitialised and complains thus:
+    
+        notes.c: In function `write_each_non_note_until':
+        notes.c:719: warning: 'cmp' might be used uninitialized in \
+            this function
+    
+    Note that gcc versions 4.1.2 and 4.4.0 do not issue this warning.
+    
+    Signed-off-by: Ramsay Jones <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit bbb1b8a35a06203020f7ab6d1ad58bcb4afe8e93
+Author: Jeff King <[email protected]>
+Date:   Mon Jun 28 04:59:07 2010 -0400
+
+    notes: check number of parameters to "git notes copy"
+    
+    Otherwise we may segfault with too few parameters.
+    
+    Signed-off-by: Jeff King <[email protected]>
+    Tested-by: Bert Wesarg <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 078e9bce1e8e2484d33d7400ae02c10954fd93ac
+Author: Ramsay Jones <[email protected]>
+Date:   Wed Jun 23 20:47:02 2010 +0100
+
+    msvc: Select the "fast" definition of the {get,put}_be32() macros
+    
+    On Intel machines, the msvc compiler defines the CPU architecture
+    macros _M_IX86 and _M_X64 (equivalent to __i386__ and __x86_64__
+    respectively). Use these macros in the pre-processor expression
+    to select the "fast" definition of the {get,put}_be32() macros.
+    
+    Signed-off-by: Ramsay Jones <[email protected]>
+    Acked-by: Jonathan Nieder <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 3f3abe3c5aa94a37bb57365262a9ad451c422e36
+Author: Michael J Gruber <[email protected]>
+Date:   Sun Jun 27 15:15:22 2010 +0200
+
+    git-rev-parse.txt: Document ":path" specifier
+    
+    The empty treeish in ":path" means "index". This is actually a special
+    case of the ":stage:path" syntax where it is documented, but mentioning
+    it also together with "treeish:path" is helpful, so do it.
+    
+    Signed-off-by: Michael J Gruber <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 492b10766f499b60bdc867c253f36d274ac51538
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:01:12 2010 -0700
+
+    Git 1.7.2-rc0
+    
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit cf4403a01070f03e5c5cc65b52c1351d20ff4e91
+Merge: 6aa2064 9f77fe0
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:07:55 2010 -0700
+
+    Merge branch 'cp/textconv-cat-file'
+    
+    * cp/textconv-cat-file:
+      git-cat-file.txt: Document --textconv
+      t/t8007: test textconv support for cat-file
+      textconv: support for cat_file
+      sha1_name: add get_sha1_with_context()
+
+commit 6aa206413ad7755577cfe29b8ffd16082c211a07
+Merge: a278aa6 64abcc4
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:07:45 2010 -0700
+
+    Merge branch 'pb/maint-perl-errmsg-no-dir'
+    
+    * pb/maint-perl-errmsg-no-dir:
+      Git.pm: better error message
+
+commit a278aa61a474eced7b676f3378dcd0f4ec5db4fc
+Merge: bcdfb20 3cae7e5
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:07:45 2010 -0700
+
+    Merge branch 'tr/send-email-8bit'
+    
+    * tr/send-email-8bit:
+      send-email: ask about and declare 8bit mails
+
+commit bcdfb20ae94138eac863b4c40263dc10f29917e5
+Merge: 4af574d 43c2325
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:07:44 2010 -0700
+
+    Merge branch 'js/maint-am-rebase-invalid-author'
+    
+    * js/maint-am-rebase-invalid-author:
+      am: use get_author_ident_from_commit instead of mailinfo when rebasing
+
+commit 4af574dbdc543b43e16838ed69890470b763136d
+Merge: a81f1a8 37d29e1
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:07:44 2010 -0700
+
+    Merge branch 'ab/blame-textconv'
+    
+    * ab/blame-textconv:
+      t/t8006: test textconv support for blame
+      textconv: support for blame
+      textconv: make the API public
+    
+    Conflicts:
+       diff.h
+
+commit a81f1a825bebe1b4d202832b804957d1547d7f60
+Merge: f526d12 5853cae
+Author: Junio C Hamano <[email protected]>
+Date:   Sun Jun 27 12:07:44 2010 -0700
+
+    Merge branch 'jn/show-num-walks'
+    
+    * jn/show-num-walks:
+      DWIM 'git show -5' to 'git show --do-walk -5'
+
+commit 52663475a97ecf15b52bee742f3c3a864f3e5fff
+Author: Michael J Gruber <[email protected]>
+Date:   Fri May 14 20:26:51 2010 +0200
+
+    t6018: make sure all tested symbolic names are different revs

     Signed-off-by: Michael J Gruber <[email protected]>
     Signed-off-by: Junio C Hamano <[email protected]>
@@ -317180,6 +317387,83 @@ Date:   Sat Sep 29 23:32:36 2007 -0700
       git-remote: exit with non-zero status after detecting errors.
       rebase -i: squash should retain the authorship of the _first_ commit
       git-add--interactive: Improve behavior on bogus input
+wasteful.  This patch fixes it as well.
+    
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 72dc52bfe6c56e37f290dd2c428d82686d6647df
+Author: Alexandre Julliard <[email protected]>
+Date:   Sat Sep 29 11:59:32 2007 +0200
+
+    git.el: Reset the permission flags when changing a file state.
+    
+    Signed-off-by: Alexandre Julliard <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 0365d885ad17031de27440fec3553675d02aa4c3
+Author: Alexandre Julliard <[email protected]>
+Date:   Sat Sep 29 11:59:07 2007 +0200
+
+    git.el: Update a file status in the git buffer upon save.
+    
+    Signed-off-by: Alexandre Julliard <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 9f5599b9829615379353f676369018c47296e1a1
+Author: Alexandre Julliard <[email protected]>
+Date:   Sat Sep 29 11:58:39 2007 +0200
+
+    git.el: Do not print a status message on every git command.
+    
+    Instead print a single message around sequences of commands that can
+    potentially take some time.
+    
+    Signed-off-by: Alexandre Julliard <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit b9b7bab4b6190ef879bb72c7fabf879fd9ca852f
+Author: Alexandre Julliard <[email protected]>
+Date:   Sat Sep 29 11:58:08 2007 +0200
+
+    git.el: Preserve file marks when doing a full refresh.
+    
+    Signed-off-by: Alexandre Julliard <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit fe50eef59b049588b7dc9b5e02707084f0516b7f
+Author: Daniel Barkalow <[email protected]>
+Date:   Sat Sep 29 19:39:54 2007 -0400
+
+    Fix adding a submodule with a remote url
+    
+    Without this, a non-path URL gets lost before the clone.
+    
+    Signed-off-by: Daniel Barkalow <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 6982ccecaf32ebf8adfcb66b4225da3b46255621
+Author: Jari Aalto <[email protected]>
+Date:   Sat Sep 29 23:34:19 2007 -0700
+
+    git-remote: exit with non-zero status after detecting error in "rm".
+    
+    Exit with non-zero status when "git remote rm" was told to
+    remove a non-existing remote.
+    
+    Signed-off-by: Jari Aalto <[email protected]>
+    Signed-off-by: Junio C Hamano <[email protected]>
+
+commit 2af89f12c666634e92cd79a6c1af6acc34104c34
+Merge: f8babc4 f4bb20c
+Author: Junio C Hamano <[email protected]>
+Date:   Sat Sep 29 23:32:36 2007 -0700
+
+    Merge branch 'maint'
+    
+    * maint:
+      git-remote: exit with non-zero status after detecting errors.
+      rebase -i: squash should retain the authorship of the _first_ commit
+      git-add--interactive: Improve behavior on bogus input
       git-add--interactive: Allow Ctrl-D to exit

 commit f4bb20cc99fe18ba0e7dd421f1d95a05c1cbbc93
@@ -483691,3 +483975,5 @@ Author: Linus Torvalds <[email protected]>
 Date:   Thu Apr 7 15:13:13 2005 -0700

     Initial revision of "git", the information manager from hell
+<<<<<   end: 6.214s
+

[lewang]

Does your PR address this issue?

[artagnon]

No, it doesn't.

[lewang]

I did something similar. My git log was 150k lines, and e-sink repeated one line randomly. I did two trials with the same result.

I only use it to look at transient data, so this probably won't get looked at in detail any time soon.