assertion fails upon execution of a valid SQL query

[monetdb-team]

Date: 2012-05-23 18:48:16 +0200
From: Babis <>
To: SQL devs <>
Version: -- development
CC: @njnes, @drstmane

Last updated: 2012-06-19 11:20:20 +0200

Comment 17277

Date: 2012-05-23 18:48:16 +0200
From: Babis <>

User-Agent: Opera/9.80 (X11; Linux x86_64; U; en) Presto/2.9.168 Version/11.52
Build Identifier: Jul2012

mclient terminates its connection to mserver5 when executing a valid SQL query such as the above:

SELECT 1, (SELECT count(*) FROM R);

Notice that a query similar to the above that swaps the projected attributes executes correctly:

SELECT (SELECT count(*) FROM R), 1;

Reproducible: Always

Steps to Reproduce:

1. CREATE TABLE R (c INT);
2. SELECT (SELECT count(*) FROM R), 1; it works fine
3. SELECT 1, (SELECT count(*) FROM R); it doesn't work

Actual Results:

sql>SELECT 1, (SELECT count() FROM R);
MAPI = (monetdb) /tmp/.s.monetdb.50000
ACTION= read_line
QUERY = SELECT 1, (SELECT count() FROM R);
ERROR = !Connection terminated

Comment 17278

Date: 2012-05-23 18:57:57 +0200
From: @drstmane

most probabaly this is actually a crash (segfault) of the server,
please check and share server console output or monetdbd (merovignian) log.

Comment 17279

Date: 2012-05-23 19:06:47 +0200
From: Babis <>

Oops, sorry, I forgot to include it:

ERR db[24300]: mserver5: ../../../MonetDB/sql/server/rel_select.c:616: rel_project: Assertion `exps_card(rel->exps) <= rel->card' failed.
MSG merovingian[9336]: database 'db' (24300) was killed by signal SIGABRT

Comment 17280

Date: 2012-05-23 20:01:36 +0200
From: @drstmane

It indeed fails with Apr2012 with assertions enabled,
but appears to work fine with assertions disabled.

We'll need to check the validity of the asssertion ...

Comment 17291

Date: 2012-05-25 21:51:53 +0200
From: @njnes

Changeset 4acfcc5693ca 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=4acfcc5693ca

Changeset description:

fixed bug #3085, ie properly keep cardinality of projection/groupby

Comment 17292

Date: 2012-05-25 21:53:49 +0200
From: @njnes

Fixed on the april2012 branch. The assertion doesn't appear anymore as the project/groupby now properly keep the cardinality (ie starting low (atom/aggr) and increased on multi row/columns)