Date: 2011-10-13 16:27:28 +0200
From: @bartscheers
To: SQL devs <>
Version: 11.5.3 (Aug2011-SP1) [obsolete]
CC: @njnes, @drstmane

Last updated: 2011-10-26 13:22:03 +0200

Comment 16405

Date: 2011-10-13 16:27:28 +0200
From: @bartscheers

User-Agent: Mozilla/5.0 (X11; Linux i686 on x86_64; rv:7.0.1) Gecko/20100101 Firefox/7.0.1
Build Identifier:

After creating and starting new database testing, and altering the
user, schema, password and authorization to testing, a (simple) UDF is created in the testing schema. Single calls work, but when called in batch simultaneous one or more TypeExceptions are thrown.

Reproducible: Always

Steps to Reproduce:

1. Create schema as in bug 10_math.sql functions not found in other than sys schema #2903.
2. Create in testing schema UD SQL function:
   CREATE FUNCTION rad(d DOUBLE) RETURNS DOUBLE BEGIN RETURN d * PI() / 180; END;
3. Init db by executing select 1;
4. %> for i in seq 50; do DOTMONETDBFILE=.testing mclient -dtesting -s"SELECT rad(55.81689);" & done

Actual Results:

Several exceptions like:
TypeException:user.s1_17[4]:'user.rad' undefined in: _7:any := user.rad(_6:dbl)
SQLException:SQLengine:Program contains errors

Expected Results:

50 times:
+------------------------+
| rad_single_value |
+========================+
| 0.97418850872349771 |
+------------------------+
1 tuple

See also bug #2903

Comment 16409

Date: 2011-10-14 08:46:17 +0200
From: @njnes

Changeset 08e5fe124329 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=08e5fe124329

Changeset description:

fix for bug #2904. We now always cleanup the function list as M5 doesn't cache
and functions anymore.

Comment 16410

Date: 2011-10-14 08:46:45 +0200
From: @njnes

fixed by always cleanup the sql function list.

Comment 16414

Date: 2011-10-14 08:56:19 +0200
From: @njnes

*** Bug #2902 has been marked as a duplicate of this bug. ***

Comment 16423

Date: 2011-10-14 10:14:28 +0200
From: @drstmane

re-opened to remind us that we should consider adding a test.

Comment 16424

Date: 2011-10-14 10:23:01 +0200
From: @grobian

Just an insight from release-engineering point of view. When this bug isn't closed/nextrelease, it won't show up (automatically) on the list of closed bugs for the next release. Since this will go unnoticed for sure, it means this bug isn't listed where it is, and with the current automation, will be reported as fixed for the release afterwards (assuming it'll be closed by then).

So how about using the Keywords field to add a status like NEEDSTEST or something, such that the bug itself can remain closed?

Comment 16425

Date: 2011-10-14 11:25:56 +0200
From: @drstmane

That would indeed be an option.

I just noticed, that closed bug with request for more information, feedback, or validation by the original reported "vanish" unnoticed, because they do no longer appear as open bugs.

in general, we would need three major categories:

• open (and that includes open but needs more info)
• fixed, but still needs feedback, info, validation, test, etc.
• closed and done.

Comment 16427

Date: 2011-10-14 12:54:05 +0200
From: @njnes

the test of both bugs is the same (the schema part doens't make a difference), ie only one needs a test.

Comment 16429

Date: 2011-10-14 13:00:57 +0200
From: @drstmane

is there a test for bug #2902 (which is marked as dublicate of this one), then?

Comment 16437

Date: 2011-10-17 14:21:54 +0200
From: @sjoerdmullender

Test was added.

Comment 16438

Date: 2011-10-17 14:27:15 +0200
From: @sjoerdmullender

Changeset f1b05d742f8f 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=f1b05d742f8f

Changeset description:

Added test for bug #2904.