Date: 2014-07-25 12:34:30 +0200
From: Martin van Dinther <<martin.van.dinther>>
To: SQL devs <>
Version: 11.17.17 (Jan2014-SP2)
CC: @njnes

Last updated: 2014-10-31 14:14:21 +0100

Comment 19928

Date: 2014-07-25 12:34:30 +0200
From: Martin van Dinther <<martin.van.dinther>>

User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:30.0) Gecko/20100101 Firefox/30.0
Build Identifier:

Calling exp(number) function for large numbers (>= 89) returns inf instead of the value as a double.

Reproducible: Always

Steps to Reproduce:

1. start mserver5
2. start mclient
3. Enter: select exp(89), log(exp(89));
   +-----------------+----------------------+
   | L1 | log_exp_single_value |
   +=================+======================+
   | inf | inf |
   +-----------------+----------------------+

Actual Results:

sql>select exp(88), log(exp(88));
+-----------------+----------------------+
| L1 | log_exp_single_value |
+=================+======================+
| 1.65163627e+38 | 88 |
+-----------------+----------------------+
1 tuple (0.894ms)

sql>select exp(89), log(exp(89));
+-----------------+----------------------+
| L1 | log_exp_single_value |
+=================+======================+
| inf | inf |
+-----------------+----------------------+
1 tuple (0.944ms)

sql>select exp(89.0), log(exp(89.0));
+-----------------+----------------------+
| L1 | log_exp_single_value |
+=================+======================+
| inf | inf |
+-----------------+----------------------+
1 tuple (0.642ms)

Expected Results:

When the result of the exp() is too large to be defined as float (which seems to be the default return type), it should return the value as a double.

Possibly exp() should always have return type double instead of float to avoid confusion.

There is a workaround for the problem by using cast( as double) to explicitly comvert the input arguments data type to a double.

sql>select exp(cast(89 as double)), log(exp(cast(89 as double)));
+--------------------------+--------------------------+
| L1 | log_exp_single_value |
+==========================+==========================+
| 4.4896128191743455e+38 | 89 |
+--------------------------+--------------------------+

Comment 19946

Date: 2014-08-04 19:29:52 +0200
From: MonetDB Mercurial Repository <>

Changeset 8f619c2ff336 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=8f619c2ff336

Changeset description:

fixed bug #3521, ie use exp(double) (and other related math functions) when
the input type isn't a real (ie single precision floating point).

Comment 19947

Date: 2014-08-04 19:30:30 +0200
From: @njnes

close we now (again) use exp(double) when the type is unknown (ie not equal to a real)

Comment 20366

Date: 2014-10-31 14:14:21 +0100
From: @sjoerdmullender

Oct2014 has been released.