Date: 2010-04-01 01:26:34 +0200
From: zeuner <>
To: SQL devs <>
Version: 11.5.3 (Aug2011-SP1) [obsolete]
CC: @njnes

Last updated: 2011-10-26 13:22:00 +0200

Comment 13167

Date: 2010-04-01 01:26:34 +0200
From: zeuner <>

I am unable to trigger this issue using the CLI which is why I attach a C
program in this report.

The symptom this Mapi error message:

ERROR = !MALException:mkey.bulk_rotate_xor_hash:Operation failed
Operands not synced on head

The issue has been verified on a 2010-03-30 (versions ..4) checkout
on an x86_64 dual core running CentOS 5.4.

Comment 13168

Date: 2010-04-08 21:19:48 +0200
From: @njnes

With your program I get (correctly) a primary key violation
error at line 993
MAPI = monetdb@localhost:50000
QUERY = INSERT
INTO transformed (base_value,general_attribute,compound,mode,activity,legacy_index,refilled,value)
SELECT gathered.base_value,general_attribute,compound,gathered.mode,state,legacy_index,refilled,value
FROM gathered,special_issue
WHERE activity=-1
AND gathered.mode=special_issue.mode
AND gathered.base_value=special_issue.base_value
MAPI = monetdb@localhost:50000
QUERY = INSERT
INTO transformed (base_value,general_attribute,compound,mode,activity,legacy_index,refilled,value)
SELECT gathered.base_value,general_attribute,compound,gathered.mode,state,legacy_index,refilled,value
FROM gathered,special_issue
WHERE activity=-1
AND gathered.mode=special_issue.mode
AND gathered.base_value=special_issue.base_value
ERROR = !SQLException:assert:INSERT INTO: PRIMARY KEY constraint 'transformed.transformed_compound_refilled_pkey' violated

Comment 13169

Date: 2010-04-09 10:49:56 +0200
From: zeuner <>

Did you by any chance use the program from [1] instead? The one from this bug only has 452 lines.

FYI, I could also reproduce it with a 2010-04-09 checkout, both on the x86_64 dual core mentioned above and on a single-cored i386 system.

[1] https://sourceforge.net/tracker/?func=detail&aid=2942744&group_id=56967&atid=482468

Comment 13170

Date: 2010-04-13 20:42:48 +0200
From: zeuner <>

Both the line number and the query SQL code show that the test was ran against the file from [1]. Therefore, the "works for me" tag applies only to [1] (which is already tagged as "fixed"), but is invalid for this bug. So I have reverted the change.

[1] http://sourceforge.net/tracker/?func=detail&aid=2942744&group_id=56967&atid=482468

Comment 13171

Date: 2010-05-04 09:32:09 +0200
From: Pseudo user for Sourceforge import <>

This bug was previously known as tracker item 2980292 at http://sourceforge.net/support/tracker.php?aid=2980292

Comment 16197

Date: 2011-09-08 13:12:32 +0200
From: @sjoerdmullender

Created attachment 70
C source to reproduce the described issue

Re-added attachment that was lost in the conversion from Sourceforge to bugzilla.

Attached file: mapi_fail.c (text/plain, 9465 bytes)
Description: C source to reproduce the described issue

Comment 16198

Date: 2011-09-08 14:00:19 +0200
From: @sjoerdmullender

Changeset 8a8d5e7e9df6 made by Sjoerd Mullender sjoerd@acm.org in the MonetDB repo, refers to this bug.

For complete details, see http//devmonetdborg/hg/MonetDB?cmd=changeset;node=8a8d5e7e9df6

Changeset description:

Test for bug #2346.
The bug was constructed using the attachment to the bug report.  It
fails as reported on the Aug2011 branch.
The stable outputs may have to be adjusted when the bug is fixed.

Comment 16337

Date: 2011-09-28 12:11:08 +0200
From: @njnes

Changeset bb23e82b4daf made by Niels Nes niels@cwi.nl in the MonetDB repo, refers to this bug.

For complete details, see http//devmonetdborg/hg/MonetDB?cmd=changeset;node=bb23e82b4daf

Changeset description:

fix bug #2346, ie return properly aligned head (ie head of input)

Comment 16338

Date: 2011-09-28 12:11:41 +0200
From: @njnes

fixed, the mkey functions now properly return aligned heads