Date: 2016-12-24 15:57:49 +0100
From: @mlkersten
To: SQL devs <>
Version: 11.17.13 (Jan2014-SP1)
CC: @njnes

Last updated: 2017-03-03 10:24:07 +0100

Comment 24852

Date: 2016-12-24 15:57:49 +0100
From: @mlkersten

User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:50.0) Gecko/20100101 Firefox/50.0
Build Identifier:

Error found by sqlsmith
assertion raised in opt_garbagecollector

Reproducible: Always

Steps to Reproduce:

select distinct 14 as c0
from sys.querylog_calls as ref_0
where ref_0.cpu is not NULL
limit 141;

Comment 24853

Date: 2016-12-24 19:31:41 +0100
From: @mlkersten

Code generation in SQL is not properly finished, as the trace below shows.
Watch for "Could not find L1.id and L1.cpu

select distinct 14 as c0 from sys.querylog_calls as ref_0 where ref_0.cpu is not NULL limit 141;
could not find L1.id
L1.cpu
=FINISHED inline 0
function user.s0_0():void;
X_0:void := querylog.define("select distinct 14 as c0 from sys.querylog_calls as ref_0 where ref_0.cpu is not null limit 141;":str, "default_pipe":str, 10:int);
sql.mvc();
calc.lng(141:lng);
calc.lng(1:lng);
(X_8:bat[:oid], X_9:bat[:timestamp], X_10:bat[:timestamp], X_11:bat[:str], X_12:bat[:lng], X_13:bat[:lng], X_14:bat[:lng], X_15:bat[:int], X_16:bat[:int]) := sql.querylog_calls();
X_17:int := calc.int(nil:int);
X_19:int := calc.int(nil:int);
X_20:bat[:oid] := algebra.select(X_15:bat[:int], X_17:int, X_19:int, true:bit, true:bit, true:bit);
algebra.projection(X_20:bat[:oid], X_15:bat[:int]);
optimizer.remap();
optimizer.costModel();
optimizer.coercions();
optimizer.evaluate();
optimizer.emptybind();
optimizer.pushselect();
optimizer.aliases();
optimizer.mitosis();
optimizer.mergetable();
optimizer.deadcode();
optimizer.aliases();
optimizer.constants();
optimizer.commonTerms();
optimizer.projectionpath();
optimizer.deadcode();
optimizer.reorder();
optimizer.matpack();
optimizer.dataflow();
optimizer.querylog();
optimizer.multiplex();
optimizer.generator();
optimizer.profiler();
optimizer.candidates();
optimizer.garbageCollector();
inline actions= 0 time=1 usec

Comment 24854

Date: 2016-12-24 20:22:47 +0100
From: @mlkersten

An freshly created database is enough.

Comment 24855

Date: 2016-12-24 20:26:01 +0100
From: @mlkersten

Another hit against the same assertion:
select distinct
57 as c0
from
sys.privileges as ref_41
left join sys.queue as ref_42
right join sys._tables as ref_43
on (ref_42.progress = ref_43.id )
right join bam.export as ref_44
on (ref_43.commit_action = ref_44.flag )
inner join sys.querylog_history as ref_45
on (ref_42.started = ref_45.defined )
on (ref_41.grantable = ref_43.id )
where ref_44.tlen is NULL
limit 78;

could not find ref_41.obj_id
ref_41.grantable
L1.started
L1.progress
ref_43.id
ref_43.commit_action
ref_44.flag
ref_44.tlen
qd.id
qd.owner
qd.defined
ql.id

Comment 25012

Date: 2017-02-12 14:14:29 +0100
From: MonetDB Mercurial Repository <>

Changeset 31c48c70f4da 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=31c48c70f4da

Changeset description:

fixed bug #6177

Comment 25013

Date: 2017-02-12 14:15:12 +0100
From: @njnes

Fixed a problem in the deadcode optimizer.

Comment 25105

Date: 2017-03-03 10:24:07 +0100
From: @sjoerdmullender

Dec2016-SP2 has been released, incorporating the fix.