DBI - Database independent interface for Perl


  use DBI;
  @data_sources = DBI->data_sources($driver_name);

  $dbh = DBI->connect($data_source, $username, $auth);
  $dbh = DBI->connect($data_source, $username, $auth, \%attr);
  $rc  = $dbh->disconnect;
  $rv  = $dbh->do($statement);
  $rv  = $dbh->do($statement, \%attr);
  $rv  = $dbh->do($statement, \%attr, @bind_values);
  $sth = $dbh->prepare($statement);
  $sth = $dbh->prepare($statement, \%attr);
  $rc = $sth->bind_col($col_num, \$col_variable);
  $rc = $sth->bind_columns(\%attr, @list_of_refs_to_vars_to_bind);

  $rv = $sth->bind_param($param_num, $bind_value);
  $rv = $sth->bind_param($param_num, $bind_value, $bind_type);
  $rv = $sth->bind_param($param_num, $bind_value, \%attr);

  $rv = $sth->execute;
  $rv = $sth->execute(@bind_values);
  @row_ary  = $sth->fetchrow_array;
  $ary_ref  = $sth->fetchrow_arrayref;
  $hash_ref = $sth->fetchrow_hashref;
  $rc = $sth->finish;
  $rv = $sth->rows;
  $rc  = $dbh->commit;
  $rc  = $dbh->rollback;

  $sql = $dbh->quote($string);
  $rc  = $h->err;
  $str = $h->errstr;
  $rv  = $h->state;


This is the DBI specification that corresponds to the DBI version 0.93 ($Date: 1998/02/13 14:27:16 $).

The DBI specification is currently evolving quite quickly so it is important to check that you have the latest copy. The RECENT CHANGES section below has a summary of user-visible changes and the Changes file supplied with the DBI holds more detailed change information.

Note also that whenever the DBI changes the drivers take some time to catch up. Recent versions of the DBI have added many new features that may not yet be supported by the drivers you use. Talk to the authors of those drivers if you need the features.

Please also read the DBI FAQ which is installed as a DBI::FAQ module so you can use perldoc to read it by executing the perldoc DBI::FAQ command.


A brief summary of significant user-visible changes in recent versions (if a recent version isn't mentioned it simply means that there were no significant user-visible changes in that version).

DBI 0.92 - 4th February 1998
Added $dbh->prepare_cached() caching variant of $dbh->prepare. Added new attributes: Active, Kids, ActiveKids, CachedKids. Added support for general-purpose 'private_' attributes.

DBI 0.91 - 10th December 1997
Fixed bug in New-style DBI->connect call which was not defaulting AutoCommit and PrintError to on.

DBI 0.86 - 16th July 1997
Added $h->{LongReadLen} and $h->{LongTruncOk} attributes for LONGs/BLOBs. Added DBI_USER and DBI_PASS env vars. See connect for usage. Added DBI->trace() to set global trace level (like per-handle $h->trace). PERL_DBI_DEBUG env var renamed DBI_TRACE (old name still works for now). Updated docs, including commit, rollback, AutoCommit and Transactions sections. Added bind_param method and execute(@bind_values) to docs.

DBI 0.85 - 25th June 1997
The 'new-style connect' (see below) now defaults to AutoCommit mode unless { AutoCommit => 0 } specified in connect attributes (see connect). New DBI_DSN env var default for connect method (supersedes DBI_DRIVER).

DBI 0.84 - 20th June 1997
Added $h->{PrintError} attribute which, if set true, causes all errors to trigger a warn(). New-style DBI->connect call now automatically sets PrintError=1 unless { PrintError => 0 } specified in the connect attributes (see connect). The old-style connect with a separate driver parameter is deprecated. Renamed $h->debug to $h->trace() and added a trace filename arg.

DBI 0.83 - 11th June 1997
Added 'new-style' driver specification syntax to the DBI->connect data_source parameter: DBI->connect( 'dbi:driver:...', $user, $passwd); The DBI->data_sources method should return data_source names with the appropriate 'dbi:driver:' prefix. DBI->connect will warn if \%attr is true but not a hash ref. Added new fetchrow methods (fetchrow_array, fetchrow_arrayref and fetchrow_hashref): Added the DBI FAQ from Alligator Descartes in module form for easy reading via ``perldoc DBI::FAQ''.


The Perl DBI is a database access Application Programming Interface (API) for the Perl Language. The DBI defines a set of functions, variables and conventions that provide a consistent database interface independant of the actual database being used.

It is important to remember that the DBI is just an interface. A thin layer of 'glue' between an application and one or more Database Drivers. It is the drivers which do the real work. The DBI provides a standard interface and framework for the drivers to operate within.

This document is a work-in-progress. Although it is incomplete it should be useful in getting started with the DBI.

Architecture of a DBI Application

             |<- Scope of DBI ->|
                  .-.   .--------------.   .-------------.
  .-------.       | |---| XYZ Driver   |---| XYZ Engine  |
  | Perl  |       |S|   `--------------'   `-------------'
  | script|  |A|  |w|   .--------------.   .-------------.
  | using |--|P|--|i|---|Oracle Driver |---|Oracle Engine|
  | DBI   |  |I|  |t|   `--------------'   `-------------'
  | API   |       |c|...
  |methods|       |h|... Other drivers
  `-------'       | |...

The API is the Application Perl-script (or Programming) Interface. The call interface and variables provided by DBI to perl scripts. The API is implemented by the DBI Perl extension.

The 'Switch' is the code that 'dispatches' the DBI method calls to the appropriate Driver for actual execution. The Switch is also responsible for the dynamic loading of Drivers, error checking/handling and other duties. The DBI and Switch are generally synonymous.

The Drivers implement support for a given type of Engine (database). Drivers contain implementations of the DBI methods written using the private interface functions of the corresponding Engine. Only authors of sophisticated/multi-database applications or generic library functions need be concerned with Drivers.

Notation and Conventions

  DBI    static 'top-level' class name
  $dbh   Database handle object
  $sth   Statement handle object
  $drh   Driver handle object (rarely seen or used in applications)
  $h     Any of the $??h handle types above
  $rc    General Return Code  (boolean: true=ok, false=error)
  $rv    General Return Value (typically an integer)
  @ary   List of values returned from the database, typically a row of data
  $rows  Number of rows processed (if available, else -1)
  $fh    A filehandle
  undef  NULL values are represented by undefined values in perl

Note that Perl will automatically destroy database and statement objects if all references to them are deleted.

Handle object attributes are shown as:


where type indicates the type of the value of the attribute (if it's not a simple scalar):

  \$   reference to a scalar: $h->{attr}       or  $a = ${$h->{attr}}
  \@   reference to a list:   $h->{attr}->[0]  or  @a = @{$h->{attr}}
  \%   reference to a hash:   $h->{attr}->{a}  or  %a = %{$h->{attr}}

General Interface Rules & Caveats

The DBI does not have a concept of a `current session'. Every session has a handle object (i.e., a $dbh) returned from the connect method and that handle object is used to invoke database related methods.

Most data is returned to the perl script as strings (null values are returned as undef). This allows arbitrary precision numeric data to be handled without loss of accuracy. Be aware that perl may not preserve the same accuracy when the string is used as a number.

Dates and times are returned as character strings in the native format of the corresponding Engine. Time Zone effects are Engine/Driver dependent.

Perl supports binary data in perl strings and the DBI will pass binary data to and from the Driver without change. It is up to the Driver implementors to decide how they wish to handle such binary data.

Multiple SQL statements may not be combined in a single statement handle, e.g., a single $sth.

Non-sequential record reads are not supported in this version of the DBI. E.g., records can only be fetched in the order that the database returned them and once fetched they are forgotten.

Positioned updates and deletes are not directly supported by the DBI. See the description of the CursorName attribute for an alternative.

Individual Driver implementors are free to provide any private functions and/or handle attributes that they feel are useful. Private driver functions can be invoked using the DBI func method. Private driver attributes are accessed just like standard attributes.

Character sets: Most databases which understand character sets have a default global charset and text stored in the database is, or should be, stored in that charset (if it's not then that's the fault of either the database or the application that inserted the data). When text is fetched it should be (automatically) converted to the charset of the client (presumably based on the locale). If a driver needs to set a flag to get that behaviour then it should do so. It should not require the application to do that.

Naming Conventions and Name Space

The DBI package and all packages below it (DBI::*) are reserved for use by the DBI. Package names beginning with DBD:: are reserved for use by DBI database drivers. All environment variables used by the DBI or DBD's begin with 'DBI_' or 'DBD_'.

The letter case used for attribute names is significant and plays an important part in the portability of DBI scripts. The case of the attribute name is used to signify who defined the meaning of that name and its values.

  Case of name  Has a meaning defined by
  ------------  ------------------------
  UPPER_CASE    Standards, e.g.,  X/Open, SQL92 etc (portable)
  MixedCase     DBI API (portable), underscores are not used.
  lower_case    Driver or Engine specific (non-portable)

It is of the utmost importance that Driver developers only use lowercase attribute names when defining private attributes. Private attribute names must be prefixed with the driver name or suitable abbreviation (e.g., ora_ for Oracle, ing_ for Ingres etc).

Data Query Methods

The DBI allows an application to `prepare' a statement for later execution. A prepared statement is identified by a statement handle object, e.g., $sth.

Typical method call sequence for a select statement:

      execute, fetch, fetch, ... finish,
      execute, fetch, fetch, ... finish,
      execute, fetch, fetch, ... finish.

Typical method call sequence for a non-select statement:


Placeholders and Bind Values

Some drivers support Placeholders and Bind Values. These drivers allow a database statement to contain placeholders, sometimes called parameter markers, that indicate values that will be supplied later, before the prepared statement is executed. For example, an application might use the following to insert a row of data into the SALES table:

  insert into sales (product_code, qty, price) values (?, ?, ?)

or the following, to select the description for a product:

  select product_description from products where product_code = ?

The ? characters are the placeholders. The association of actual values with placeholders is known as binding and the values are referred to as bind values. Undefined values or undef can be used to indicate null values.

Without using placeholders, the insert statement above would have to contain the literal values to be inserted and it would have to be re-prepared and re-executed for each row. With placeholders, the insert statement only needs to be prepared once. The bind values for each row can be given to the execute method each time it's called. By avoiding the need to re-prepare the statement for each row the application typically many times faster! Here's an example:

  my $sth = $dbh->prepare(q{
    insert into sales (product_code, qty, price) values (?, ?, ?)
  }) || die $dbh->errstr;
  while(<>) {
      my($product_code, $qty, $price) = split(/,/);
      $sth->execute($product_code, $qty, $price) || die $dbh->errstr;
  $dbh->commit || die $dbh->errstr;

See execute and bind_param for more details.

See bind_column for a related method used to associate perl variables with the output columns of a select statement.

SQL - A Query Language

Most DBI drivers require applications to use a dialect of SQL (the Structured Query Language) to interact with the database engine. These links may provide some useful information about SQL:

The DBI itself does not mandate or require any particular language to be used. It is language independant. In ODBC terms it is always in pass-thru mode. The only requirement is that queries and other statements must be expressed as a single string of letters passed as the first argument to the prepare method.


DBI Class Methods

  $dbh = DBI->connect($data_source, $username, $password);
  $dbh = DBI->connect($data_source, $username, $password, \%attr);

Establishes a database connection (session) to the requested data_source. Returns a database handle object.

Multiple simultaneous connections to multiple databases through multiple drivers can be made via the DBI. Simply make one connect call for each and keep a copy of each returned database handle.

The $data_source value should begin with 'dbi:driver_name:'. That prefix will be stripped off and the driver_name part is used to specify the driver (letter case is significant). As a convenience, if the $data_source field is undefined or empty the DBI will substitute the value of the environment variable DBI_DSN if any.

If driver is not specified, the environment variable DBI_DRIVER is used. If that variable is not set then the connect dies.

Examples of $data_source values:


There is no standard for the text following the driver name. Each driver is free to use whatever syntax it wants. The only requirement the DBI makes is that all the information is supplied in a single string. You must consult the documentation for the drivers you are using for a description of the syntax they require. (Where a driver author needs to define a syntax for the data_source it is recommended that they follow the ODBC style, the last example above.)

If $username or $password are undefined (rather than empty) then the DBI will substitute the values of the DBI_USER and DBI_PASS environment variables respectively. The use of the environment for these values is not recommended for security reasons. The mechanism is only intended to simplify testing.

DBI->connect automatically installs the driver if it has not been installed yet. Driver installation always returns a valid driver handle or it dies with an error message which includes the string 'install_driver' and the underlying problem. So, DBI->connect will die on a driver installation failure and will only return undef on a connect failure, for which $DBI::errstr will hold the error.

The $data_source argument (with the 'dbi:...:' prefix removed) and the $username and $password arguments are then passed to the driver for processing. The DBI does not define any interpretation for the contents of these fields. The driver is free to interpret the data_source, username and password fields in any way and supply whatever defaults are appropriate for the engine being accessed (Oracle, for example, uses the ORACLE_SID and TWO_TASK env vars if no data_source is specified).

The AutoCommit and PrintError attributes for each connection default to default to on (see AutoCommit and PrintError for more information).

The \%attr parameter can be used to alter the default settings of the PrintError, RaiseError and AutoCommit attributes. For example:

  $dbh = DBI->connect($data_source, $user, $pass, {
        PrintError => 0,
        AutoCommit => 0

These are currently the only defined uses for the DBI->connect \%attr.

Portable applications should not assume that a single driver will be able to support multiple simultaneous sessions.

Where possible each session ($dbh) is independent from the transactions in other sessions. This is useful where you need to hold cursors open across transactions, e.g., use one session for your long lifespan cursors (typically read-only) and another for your short update transactions.

For compatibility with old DBI scripts the driver can be specified by passing its name as the fourth argument to connect (instead of \%attr):

  $dbh = DBI->connect($data_source, $user, $pass, $driver);

In this 'old-style' form of connect the $data_source should not start with 'dbi:driver_name:' and, even if it does, the embedded driver_name will be ignored. The $dbh->{AutoCommit} attribute is undefined. The $dbh->{PrintError} attribute is off. And the old DBI_DBNAME env var is checked if DBI_DSN is not defined.

  @ary = DBI->available_drivers;
  @ary = DBI->available_drivers($quiet);

Returns a list of all available drivers by searching for DBD::* modules through the directories in @INC. By default a warning will be given if some drivers are hidden by others of the same name in earlier directories. Passing a true value for $quiet will inhibit the warning.

  @ary = DBI->data_sources($driver);

Returns a list of all data sources (databases) available via the named driver. The driver will be loaded if not already. If $driver is empty or undef then the value of the DBI_DRIVER environment variable will be used.

Note that many drivers have no way of knowing what data sources might be available for it and thus, typically, return an empty list.

  DBI->trace($trace_level, $trace_file)

DBI trace information can be enabled for all handles using this DBI class method. To enable trace information for a specific handle use the similar $h->trace method described elsewhere.

Use $trace_level 2 to see detailed call trace information including parameters and return values. The trace output is detailed and typically very useful.

Use $trace_level 0 to disable the trace.

If $trace_filename is specified then the file is opened in append mode and all trace output (including that from other handles) is redirected to that file.

See also the $h->trace() method and DEBUGGING for information about the DBI_TRACE environment variable.

DBI Utility Functions

  $str = DBI::neat($value, $maxlen);

Return a string containing a neat (and tidy) representation of the supplied value.

Strings will be quoted (but internal quotes will not be escaped). Numeric values will usually be unquoted. Undefined (NULL) values will be shown as undef (without quotes). Unprintable characters will be replaced by dot (.). For result strings longer than $maxlen (0 or undef defaults to 400 characters) the result string will be truncated to $maxlen-4 and ...' will be appended.

This function is designed to format values for human consumption. It is used internally by the DBI for trace output. It should typically not be used for formating values for database use.

  $str = DBI::neat_list(\@listref, $maxlen, $field_sep);

Calls DBI::neat on each element of the list and returns a string containing the results joined with $field_sep. $field_sep defaults to ", ".

  $rows = DBI::dump_results($sth, $maxlen, $lsep, $fsep, $fh);

Fetches all the rows from $sth, calls DBI::neat_list for each row and prints the results to $fh (defaults to STDOUT) separated by $lsep (default "\n"). $fsep defaults to ", " and $maxlen defaults to 35.

This function is designed as a handy utility for prototyping and testing queries. Since it uses neat_list which uses neat which formats the string for reading by humans, it's not recomended for data transfer applications.

DBI Dynamic Attributes

These attributes are always associated with the last handle used.

Where an attribute is equivalent to a method call, then refer to the method call for all related documentation.

Warning: these attributes are provided as a convenience but they do have limitations. Specifically, because they are associated with the last handle used, they should only be used immediately after calling the method which 'sets' them. They have a 'short lifespan'. There may also be problems with the multi-threading in 5.005.

If in any doubt, use the corresponding method call.

Equivalent to $h->err.

Equivalent to $h->errstr.

Equivalent to $h->state.

Equivalent to $h->rows.


  $rv = $h->err;

Returns the native database engine error code from the last driver function called.

  $str = $h->errstr;

Returns the native database engine error message from the last driver function called.

  $str = $h->state;

Returns an error code in the standard SQLSTATE five character format. Note that the specific success code 00000 is translated to (false). If the driver does not support SQLSTATE then state will return S1000 (General Error) for all errors.

  $h->trace($trace_level, $trace_filename);

DBI trace information can be enabled for a specific handle (and any future children of that handle) by setting the trace level using the trace method.

Use $trace_level 2 to see detailed call trace information including parameters and return values. The trace output is detailed and typically very useful.

Use $trace_level 0 to disable the trace.

If $trace_filename is specified then the file is opened in append mode and all trace output (including that from other handles) is redirected to that file.

See also the DBI->trace() method and DEBUGGING for information about the DBI_TRACE environment variable.

  $h->func(@func_arguments, $func_name);

The func method can be used to call private non-standard and non-portable methods implemented by the driver. Note that the function name is given as the last argument.

This method is not directly related to calling stored procedures. Calling stored procedures is currently not defined by the DBI. Some drivers, such as DBD::Oracle, support it in non-portable ways. See driver documentation for more details.


These attributes are common to all types of DBI handles.

Some attributes are inherited by child handles. That is, the value of an inherited attribute in a newly created statement handle is the same as the value in the parent database handle. Changes to attributes in the new statement handle do not affect the parent database handle and changes to the database handle do not affect existing statement handles, only future ones.

Attempting to set or get the value of an unknown attribute is fatal, except for private driver specific attributes (which all have names starting with a lowercase letter).


  $h->{AttributeName} = ...;    # set/write
  ... = $h->{AttributeName};    # get/read

Warn (boolean, inherited)
Enables useful warnings for certain bad practices. Enabled by default. Some emulation layers, especially those for perl4 interfaces, disable warnings.

Active (boolean, read-only)
True if the handle object is 'active'. This is rarely used in applications. The exact meaning of active is somewhat vague at the moment. For a database handle it typically means that the handle is connected to a database ($dbh->disconnect should set Active off). For a statement handle it typically means that the handle is a select that may have more data to fetch ($dbh->finish or fetching all the data should set Active off).

Kids (integer, read-only)
For a driver handle, Kids is the number of currently existing database handles that were created from that driver handle. For a database handle, Kids is the number of currently existing statement handles that were created from that database handle.

ActiveKids (integer, read-only)
Like Kids (above), but only counting those that are Active (as above).

CachedKids (hash ref)
For a database handle, returns a reference to the cache (hash) of statement handles created by the prepare_cached method. For a driver handle, it would return a reference to the cache (hash) of statement handles created by the (not yet implemented) connect_cached method.

CompatMode (boolean, inherited)
Used by emulation layers (such as Oraperl) to enable compatible behaviour in the underlying driver (e.g., DBD::Oracle) for this handle. Not normally set by application code.

InactiveDestroy (boolean)
This attribute can be used to disable the effect of destroying a handle (which would normally close a prepared statement or disconnect from the database etc). It is specifically designed for use in unix applications which 'fork' child processes. Either the parent or the child process, but not both, should set InactiveDestroy on all their handles. This attribute does not disable an explicit call to the disconnect method.

PrintError (boolean, inherited)
This attribute can be used to force errors to generate warnings (using warn) in addition to returning error codes in the normal way. When set on, any method which results in an error occuring ($DBI::err being set true) will cause the DBI to effectively do warn(``$DBI::errstr''). Note that the contents of the warning are currently just $DBI::errstr but that may change and should not be relied upon.

By default DBI->connect sets PrintError on (except for old-style connect usage, see connect for more details).

If desired, the warnings can be caught and processed using a $SIG{__WARN__} handler or modules like CGI::ErrorWrap.

RaiseError (boolean, inherited)
This attribute can be used to force errors to raise exceptions rather than simply return error codes in the normal way. It defaults to off. When set on, any method which results in an error occuring ($DBI::err being set true) will cause the DBI to effectively do croak(``$DBI::errstr'').

If PrintError is also on then the PrintError is done before the RaiseError unless no __DIE__ handler has been defined, in which case PrintError is skipped since the croak will print the message.

Note that the contents of $@ are currently just $DBI::errstr but that may change and should not be relied upon.

ChopBlanks (boolean, inherited)
This attribute can be used to control the trimming of trailing space characters from fixed width char fields. No other field types are affected.

The default is false (it is possible that that may change). Applications that need specific behaviour should set the attribute as needed. Emulation interfaces should set the attribute to match the behaviour of the interface they are emulating.

Drivers are not required to support this attribute but any driver which does not must arrange to return undef as the attribute value.

LongReadLen (integer, inherited)
This attribute may be used to control the maximum length of 'long' (or 'blob') fields which the driver will read from the database automatically when it fetches each row of data. A value of 0 means don't automatically fetch any long data (fetch should return undef for long fields when LongReadLen is 0).

The default is typically 0 (zero) bytes but may vary between drivers. Most applications using long fields will set this value to slightly larger than the longest long field value which will be fetched.

Changing the value of LongReadLen for a statement handle after it's been prepare()'d will typically have no effect so it's usual to set LongReadLen on the $dbh before calling prepare.

The LongReadLen attribute only relates to fetching/reading long values it is not involved in inserting/updating them.

See LongTruncOk about truncation behaviour.

LongTruncOk (boolean, inherited)
This attribute may be used to control the effect of fetching a long field value which has been truncated (typically because it's longer than the value of the LongReadLen attribute).

By default LongTruncOk is false and fetching a truncated long value will cause the fetch to fail. (Applications should always take care to check for errors after a fetch loop in case an error, such as a divide by zero or long field truncation, caused the fetch to terminate prematurely.)

If a fetch fails due to a long field truncation when LongTruncOk is false, many drivers will allow you to continue fetching further rows.

The DBI provides a way to store extra information in a DBI handle as 'private' attributes. The DBI will allow you to store and retreive any attribute which has a name starting with 'private_'. It is strongly recommended that you use just once private attribute (e.g., use a hash ref) and give it a name that includes the module or application that the attribute relates to (e.g., 'private_DBD_Foo_thingy').


Database Handle Methods

  $sth = $dbh->prepare($statement)          || die $dbh->errstr;
  $sth = $dbh->prepare($statement, \%attr)  || die $dbh->errstr;

Prepare a single statement for execution by the database engine and return a reference to a statement handle object which can be used to get attributes of the statement and invoke the execute method.

Note that prepare should never execute a statement, even if it is not a select statement, it only prepares it for execution. (Having said that, some drivers, notably Oracle, will execute data definition statements such as create/drop table when they are prepared. In practice this is rarely a problem.)

Drivers for engines which don't have the concept of preparing a statement will typically just store the statement in the returned handle and process it when $sth->execute is called. Such drivers are likely to be unable to give much useful information about the statement, such as $sth->{NUM_OF_FIELDS}, until after $sth->execute has been called. Portable applications should take this into account.

  $sth = $dbh->prepare_cached($statement)          || die $dbh->errstr;
  $sth = $dbh->prepare_cached($statement, \%attr)  || die $dbh->errstr;

Like prepare except that the statement handled returned will be stored in a hash associated with the $dbh. If another call is made to prepare_cached with the same parameter values then the corresponding cached $sth will be returned (and the database server will not be contacted).

This cacheing can be useful in some applications but it can also cause problems and should be used with care. Currently a warning will be generated if the cached $sth being returned is active (finish has not been called on it).

The cache can be accessed (and cleared) via the CachedKids attribute.

  $rc  = $dbh->do($statement)           || die $dbh->errstr;
  $rc  = $dbh->do($statement, \%attr)   || die $dbh->errstr;
  $rv  = $dbh->do($statement, \%attr, @bind_values) || ...

Prepare and execute a statement. Returns the number of rows affected (-1 if not known or not available) or undef on error.

This method is typically most useful for non-select statements which either cannot be prepared in advance (due to a limitation in the driver) or which do not need to be executed repeatedly.

The default do method is logically similar to:

  sub do {
      my($dbh, $statement, $attr, @bind_values) = @_;
      my $sth = $dbh->prepare($statement) or return undef;
      $sth->execute(@bind_values) or return undef;
      my $rows = $sth->rows;
      ($rows == 0) ? "0E0" : $rows;


  my $rows_deleted = $dbh->do(q{
      delete from table
      where status = 'DONE'
  }) || die $dbh->errstr;

Using placeholders and @bind_values with the do method can be useful because it avoids the need to correctly quote any variables in the $statement.

  $rc  = $dbh->commit     || die $dbh->errstr;

Commit (make permanent) the most recent series of database changes if the database supports transactions.

If the database supports transactions and AutoCommit is on then the commit should issue a ``commit ineffective with AutoCommit'' warning.

See also Transactions.

  $rc  = $dbh->rollback   || die $dbh->errstr;

Roll-back (undo) the most recent series of uncommitted database changes if the database supports transactions.

If the database supports transactions and AutoCommit is on then the rollback should issue a ``rollback ineffective with AutoCommit'' warning.

See also Transactions.

  $rc  = $dbh->disconnect   || warn $dbh->errstr;

Disconnects the database from the database handle. Typically only used before exiting the program. The handle is of little use after disconnecting.

The transaction behaviour of the disconnect method is, sadly, undefined. Some database systems (such as Oracle and Ingres) will automatically commit any outstanding changes, but others (such as Informix) will rollback any outstanding changes. Applications should explicitly call commit or rollback before calling disconnect.

The database is automatically disconnected (by the DESTROY method) if still connected when there are no longer any references to the handle. The DESTROY method for each driver should explicitly call rollback to undo any uncommitted changes. This is vital behaviour to ensure that incomplete transactions don't get committed simply because Perl calls DESTROY on every object before exiting.

If you disconnect from a database while you still have active statement handles you will get a warning. The statement handles should either be cleared (destroyed) before disconnecting or the finish method called on each one.

  $rc = $dbh->ping;

Attempts to determine, in a reasonably efficient way, if the database server is still running and the connection to it is still working.

The default implementation currently always returns true without actually doing anything. Individual drivers should implement this function in the most suitable manner for their database engine.

Very few applications would have any use for this method. See the specialist Apache::DBI module for one example usage.

  $sql = $dbh->quote($string);

Quote a string literal for use in an SQL statement by escaping any special characters (such as quotation marks) contained within the string and adding the required type of outer quotation marks.

  $sql = sprintf "select foo from bar where baz = %s",

For most database types quote would return 'Don''t' (including the outer quotation marks).

An undefined $string value will be returned as NULL (without quotation marks).

Quote may not be able to deal with all possible input (such as binary data) and should not be relied upon for security.

Database Handle Attributes

This section describes attributes specific to database handles.

Changes to these database handle attributes do not affect any other existing or future database handles.

Attempting to set or get the value of an unknown attribute is fatal, except for private driver specific attributes (which all have names starting with a lowercase letter).


  $h->{AutoCommit} = ...;       # set/write
  ... = $h->{AutoCommit};       # get/read

AutoCommit (boolean)
If true then database changes cannot be rolled-back (undone). If false then database changes automatically occur within a 'transaction' which must either be committed or rolled-back using the commit or rollback methods.

Drivers should always default to AutoCommit mode. (An unfortunate choice forced on the DBI by ODBC and JDBC conventions.)

Attempting to set AutoCommit to an unsupported value is a fatal error. This is an important feature of the DBI. Applications which need full transaction behaviour can set $dbh->{AutoCommit}=0 (or via connect) without having to check the value was assigned okay.

For the purposes of this description we can divide databases into three categories:

  Database which don't support transactions at all.
  Database in which a transaction is always active.
  Database in which a transaction must be explicitly started ('BEGIN WORK').

* Database which don't support transactions at all

For these databases attempting to turn AutoCommit off is a fatal error. Commit and rollback both issue warnings about being ineffective while AutoCommit is in effect.

* Database in which a transaction is always active

These are typically mainstream commercial relational databases with 'ANSI standandard' transaction behaviour.

If AutoCommit is off then changes to the database won't have any lasting effect unless commit is called (but see also disconnect). If rollback is called then any changes since the last commit are undone.

If AutoCommit is on then the effect is the same as if the DBI were to have called commit automatically after every successful database operation. In other words, calling commit or rollback explicitly while AutoCommit is on would be ineffective because the changes would have already been commited.

Changing AutoCommit from off to on should issue a commit in most drivers.

Changing AutoCommit from on to off should have no immediate effect.

For databases which don't support a specific auto-commit mode, the driver has to commit each statement automatically using an explicit COMMIT after it completes successfully (and roll it back using an explicit ROLLBACK if it fails). The error information reported to the application will correspond to the statement which was executed, unless it succeeded and the commit or rollback failed.

* Database in which a transaction must be explicitly started

For these database the intention is to have them act like databases in which a transaction is always active (as described above).

To do this the DBI driver will automatically begin a transaction when AutoCommit is turned off (from the default on state) and will automatically begin another transaction after a commit or rollback.

In this way, the application does not have to treat these databases as a special case.


Statement Handle Methods

  $rc = $sth->bind_param($param_num, $bind_value)  || die $sth->errstr;
  $rv = $sth->bind_param($param_num, $bind_value, \%attr)     || ...
  $rv = $sth->bind_param($param_num, $bind_value, $bind_type) || ...

The bind_param method can be used to bind (assign/associate) a value with a placeholder embedded in the prepared statement. Placeholders are indicated with question mark character (?). For example:

  $dbh->{RaiseError} = 1;        # save having to check each method call
  $sth = $dbh->prepare("select name, age from people where name like ?");
  $sth->bind_param(1, "John%");  # placeholders are numbered from 1

Note that the ? is not enclosed in quotation marks even when the placeholder represents a string. Some drivers also allow :1, :2 etc and :name style placeholders in addition to ? but their use is not portable.

Some drivers do not support placeholders.

With most drivers placeholders can't be used for any element of a statement that would prevent the database server validating the statement and creating a query execution plan for it. For example:

  "select name, age from ?"         # wrong
  "select name, ?   from people"    # wrong

Also, placeholders can only represent single scalar values, so this statement, for example, won't work as expected for more than one value:

  "select name, age from people where name in (?)"    # wrong

The \%attr parameter can be used to specify the data type the placeholder should have. Typically the driver is only interested in knowing if the placeholder should be bound as a number or a string.

  $sth->bind_param(1, $value, { TYPE => SQL_INTEGER });

As a short-cut for this common case, the data type can be passed directly inplace of the attr hash reference. This example is equivalent to the one above:

  $sth->bind_param(1, $value, SQL_INTEGER);

The TYPE cannot be changed after the first bind_param call (but it can be left unspecified, in which case it defaults to the previous value).

Perl only has string and number scalar data types. All database types that aren't numbers are bound as strings and must be in a format the database will understand.

Undefined values or undef are be used to indicate null values.

  $rc = $sth->bind_param_inout($param_num, \$bind_value, $max_len)  || die $sth->errstr;
  $rv = $sth->bind_param_inout($param_num, \$bind_value, $max_len, \%attr)     || ...
  $rv = $sth->bind_param_inout($param_num, \$bind_value, $max_len, $bind_type) || ...

This method acts like bind_param but also enables values to be output from (updated by) the statement. (The statement is typically a call to a stored procedure.). The $bind_value must be passed as a reference to the actual value to be used.

The additional $max_len parameter specifies the amount of memory to allocate to $bind_value for the new value. Truncation behaviour, if the value is longer than $max_len, is currently undefined.

It is expected that few drivers will support this method. The only driver currently known to do so is DBD::Oracle. It should not be used for database independent applications.

  $rv = $sth->execute                || die $sth->errstr;
  $rv = $sth->execute(@bind_values)  || die $sth->errstr;

Perform whatever processing is necessary to execute the prepared statement. An undef is returned if an error occurs, a successful execute always returns true (see below). It is always important to check the return status of execute (and most other DBI methods).

For a non-select statement execute returns the number of rows affected (if known). Zero rows is returned as ``0E0'' which Perl will treat as 0 but will regard as true. If the number of rows affected is not known then execute returns -1.

For select statements execute simply 'starts' the query within the Engine. Use one of the fetch methods to retreive the data after calling execute. The execute method does not return the number of rows that will be returned by the query (because most Engines can't tell in advance), it simply returns a true value.

If any arguments are given then execute will effectively call bind_param for each value before executing the statement. Values bound in this way are usually treated as SQL_VARCHAR types unless the driver can determine the correct type (which is rare) or bind_param (or bind_param_inout) has already been used to specify the type.

  $ary_ref = $sth->fetchrow_arrayref;
  $ary_ref = $sth->fetch;    # alias

Fetches the next row of data and returns a reference to an array holding the field values. If there are no more rows fetchrow_arrayref returns undef. Null values are returned as undef. This is the fastest way to fetch data, particularly if used with $sth->bind_columns.

Note that the same hash ref will be returned for each fetch so don't store the ref and then use it after a later fetch.

 @ary = $sth->fetchrow_array;

An alternative to fetchrow_arrayref. Fetches the next row of data and returns it as an array holding the field values. If there are no more rows fetchrow_array returns an empty list. Null values are returned as undef.

 $hash_ref = $sth->fetchrow_hashref;

An alternative to fetchrow_arrayref. Fetches the next row of data and returns it as a reference to a hash containing field name and field value pairs. Null values are returned as undef. If there are no more rows fetchhash returns undef.

The keys of the hash are the same names returned by $sth->{NAME}. If more than one field has the same name there will only be one entry in the returned hash.

Because of the extra work fetchrow_hashref and perl have to perform it is not as efficient as fetchrow_arrayref or fetchrow_array and is not recommended where performance is very important. Currently a new hash reference is returned for each row. This is likely to change in the future so don't rely on it.

  $tbl_ary_ref = $sth->fetchall_arrayref;

The fetchall_arrayref method can be used to fetch all the data to be returned from a prepared statement. It returns a reference to an array which contains one array reference per row (as returned by fetchrow_arrayref).

If there are no rows to return, fetchall_arrayref returns a reference to an empty array.

  $rc  = $sth->finish;

Indicates that no more data will be fetched from this statement before it is either prepared again or destroyed. It can sometimes be helpful to call this method where appropriate in order to allow the server to free up any internal resources (such as read locks) currently being held. It does not affect the transaction status of the session.

The finish method has nothing to do with transactions. It's mostly an internal 'housekeeping' method. There's no need to call finish if you're about to destroy or re-use the statement handle. See also disconnect.

  $rv = $sth->rows;

Returns the number of rows affected by the last database altering command, or -1 if not known or not available.

Generally you can only rely on a row count after a do or non-select execute (for some specific operations like update and delete) or after fetching all the rows of a select statement.

For select statements it is generally not possible to know how many rows will be returned except by fetching them all. Some drivers will return the number of rows the application has fetched so far but others may return -1 until all rows have been fetched.

  $rc = $sth->bind_col($column_number, \$var_to_bind);
  $rc = $sth->bind_col($column_number, \$var_to_bind, \%attr);

Binds an output column (field) of a select statement to a perl variable. Whenever a row is fetched from the database the corresponding perl variable is automatically updated. There is no need to fetch and assign the values manually. This makes using bound variables very efficient. See bind_columns below for an example. Note that column numbers count up from 1.

The binding is performed at a very low level using perl aliasing so there is no extra copying taking place. So long as the driver uses the correct internal DBI call to get the array the fetch function returns, it will automatically support column binding.

The bind_param method performs a similar function for input variables. See also Placeholders and Bind Values for more information.

  $rc = $sth->bind_columns(\%attr, @list_of_refs_to_vars_to_bind);


  $dbh->{RaiseError} = 1; # do this, or check every call for errors
  $sth = $dbh->prepare(q{ select region, sales from sales_by_region });
  my($region, $sales);
  # Bind perl variables to columns.
  $rv = $sth->bind_columns(undef, \$region, \$sales);
  # you can also use perl's \(...) syntax (see perlref docs):
  #     $sth->bind_columns(undef, \($region, $sales));
  # Column binding is the most efficient way to fetch data
  while($sth->fetch) {
      print "$region: $sales\n";

Calls bind_col for each column of the select statement. bind_columns will croak if the number of references does not match the number of fields.

Statement Handle Attributes

This section describes attributes specific to statement handles. Most of these attributes are read-only.

Changes to these statement handle attributes do not affect any other existing or future statement handles.

Attempting to set or get the value of an unknown attribute is fatal, except for private driver specific attributes (which all have names starting with a lowercase letter).


  ... = $h->{NUM_OF_FIELDS};    # get/read

Note that some drivers cannot provide valid values for some or all of these attributes until after $sth->execute has been called.

NUM_OF_FIELDS (integer, read-only)
Number of fields (columns) the prepared statement will return. Non-select statements will have NUM_OF_FIELDS == 0.

NUM_OF_PARAMS (integer, read-only)
The number of parameters (placeholders) in the prepared statement. See SUBSTITUTION VARIABLES below for more details.

NAME (array-ref, read-only)
Returns a reference to an array of field names for each column. The names may contain spaces but should not be truncated or have any trailing space.

  print "First column name: $sth->{NAME}->[0]\n";

NULLABLE (array-ref, read-only)
Returns a reference to an array indicating the possibility of each column returning a null.

  print "First column may return NULL\n" if $sth->{NULLABLE}->[0];

CursorName (string, read-only)
Returns the name of the cursor associated with the statement handle if available. If not available or the database driver does not support the "where current of ..." SQL syntax then it returns undef.



Transactions are a fundamental part of any robust database system. They protect against errors and database corruption by ensuring that sets of related changes to the database take place in atomic (indivisible, all-or-nothing) units.

See AutoCommit for details of using AutoCommit with various types of database.

This section applies to databases which support transactions and where AutoCommit is off.

The recommended way to implement robust transactions in Perl applications is to make use of eval { ... } (which is very fast, unlike eval "...").

  eval {
      foo(...)   # do lots of work here
      bar(...)   # including inserts
      baz(...)   # and updates
  if ($@) {
      # add other application on-error-clean-up code here
  else {

The code in foo(), or any other code executed from within the curly braces, can be implemented in this way:

  $h->method(@args) || die $h->errstr

or the $h->{RaiseError} attribute can be set on, in which case the DBI will automatically croak() on error so you don't have to test the return value of each method call. See RaiseError for more details.

A major advantage of the eval approach is that the transaction will be properly rolled back if any code in the inner application croaks or dies for any reason. The major advantage of using the $h->{RaiseError} attribute is that all DBI calls will be checked automatically. Both techniques are recommended.

Handling BLOB / Long / Memo Fields

Many databases support 'blob' (binary large objects), 'long' or similar datatypes for holding very long strings or large amounts of binary data in a single field. Some databases support variable length long values over 2,000,000,000 bytes in length.

Since values of that size can't usually be held in memory and because databases can't usually know in advance the length of the longest long that will be returned from a select statement (unlike other data types) some special handling is required.

In this situation the value of the $h->{LongReadLen} attribute is used to determine how much buffer space to allocate for such fields. The $h->{LongTruncOk} attribute is used to determine how to behave if a fetched value can't fit into the buffer.

Simple Example

  my $dbh = DBI->connect("dbi:Oracle:$data_source", $user, $password)
      || die "Can't connect to $data_source: $DBI::errstr";

  my $sth = $dbh->prepare( q{
          SELECT name, phone
          FROM mytelbook
  }) || die "Can't prepare statement: $DBI::errstr";

  my $rc = $sth->execute
      || die "Can't execute statement: $DBI::errstr";

  print "Query will return $sth->{NUM_OF_FIELDS} fields.\n\n";

  print "$sth->{NAME}->[0]: $sth->{NAME}->[1]\n";
  while (($name, $phone) = $sth->fetchrow_array) {
      print "$name: $phone\n";
  # check for problems which may have terminated the fetch early
  warn $DBI::errstr if $DBI::err;



In addition to the trace method you can enable the same trace information by setting the DBI_TRACE environment variable before starting perl.

On unix-like systems using a bourne-like shell you can do this easily for a single command:

  DBI_TRACE=2 perl

If DBI_TRACE is set to a non-numeric value then it is assumed to be a file name and the trace level will be set to 2 with all trace output will be appended to that file.

See also the trace method.


(This section needs more words about causes and remedies.)

Fatal Errors

DBI/DBD internal version mismatch (DBI is v%d/s%d, DBD %s expected v%d/s%d) you need to rebuild either the DBD driver or the DBI
DBD driver has not implemented the AutoCommit attribute
Can't [sg]et %s->{%s}: unrecognised attribute
panic: DBI active kids (%d) > kids (%d)
panic: DBI active kids (%d) < 0 or > kids (%d)


DBI Handle cleared whilst still holding %d cached kids!
DBI Handle cleared whilst still active!
DBI Handle has uncleared implementors data
DBI Handle has %d uncleared child handles


Database Documentation

SQL Language Reference Manual.

Books and Journals

 Programming Perl 2nd Ed. by Larry Wall, Tom Christiansen & Randal Schwartz.
 Learning Perl by Randal Schwartz.

 Dr Dobb's Journal, November 1996.
 The Perl Journal, April 1997.

Manual Pages

perl(1), perlmod(1), perlbook(1)

Mailing List

The dbi-users mailing list is the primary means of communication among uses of the DBI and its related modules. Subscribe and unsubscribe via:

Mailing list archives are held at:

Assorted Related WWW Links

The DBI 'Home Page' (not maintained by me):

Other related links:


Please also read the DBI FAQ which is installed as a DBI::FAQ module so you can use perldoc to read it by executing the perldoc DBI::FAQ command.


DBI by Tim Bunce. This pod text by Tim Bunce, J. Douglas Dunlop, Jonathan Leffler and others. Perl by Larry Wall and the perl5-porters.


The DBI module is Copyright (c) 1995,1996,1997 Tim Bunce. England. All rights reserved.

You may distribute under the terms of either the GNU General Public License or the Artistic License, as specified in the Perl README file.


I would like to acknowledge the valuable contributions of the many people I have worked with on the DBI project, especially in the early years (1992-1994). In no particular order: Kevin Stock, Buzz Moschetti, Kurt Andersen, Ted Lemon, William Hails, Garth Kennedy, Michael Peppler, Neil S. Briscoe, Jeff Urlwin, David J. Hughes, Jeff Stander, Forrest D Whitcher, Larry Wall, Jeff Fried, Roy Johnson, Paul Hudson, Georg Rehfeld, Steve Sizemore, Ron Pool, Jon Meek, Tom Christiansen, Steve Baumgarten, Randal Schwartz, and a whole lot more.



Commercial support for Perl and the DBI, DBD::Oracle and Oraperl modules can be arranged via The Perl Clinic. See for more details.


        data types (ISO type numbers and type name conversions)
        error handling
        data dictionary methods
        test harness support methods


See the DBI FAQ for a more comprehensive list of FAQs. Use the perldoc DBI::FAQ command to read it.

How fast is the DBI?

To measure the speed of the DBI and DBD::Oracle code I modified DBD::Oracle such that you can set an attribute which will cause the same row to be fetched from the row cache over and over again (without involving Oracle code but exercising *all* the DBI and DBD::Oracle code in the code path for a fetch).

The results (on my lightly loaded Sparc 10) fetching 50000 rows using:

        1 while $csr->fetch;

were: one field: 5300 fetches per cpu second (approx) ten fields: 4000 fetches per cpu second (approx)

Obviously results will vary between platforms but it does give a feel for the current theoretical maximum performance. By way of comparison, using the code:

        1 while @row = $csr->fetchrow_array;

(fetchrow_array is roughly the same as ora_fetch) gives:

        one field:   3100 fetches per cpu second (approx)
        ten fields:  1000 fetches per cpu second (approx)

Notice the slowdown and the more dramatic impact of extra fields. (The fields were all one char long. The impact would be even bigger for longer strings.)

Changing that slightly to represent actually _doing_ something in perl with the fetched data:

    while(@row = $csr->fetchrow_array) {
        $hash{++$i} = [ @row ];

gives: ten fields: 500 fetches per cpu second (approx)

That simple addition has *halved* the performance.

I therefore conclude that DBI and DBD::Oracle overheads are small compared with Perl language overheads (and probably database overheads).

So, if you think the DBI or your driver is slow, try replacing your fetch loop with just:

        1 while $csr->fetch;

and time that. If that doesn't help much then point the finger at the database, the platform, the network etc. But think carefully before pointing it at the DBI or your driver.

(Having said all that, if anyone can show me how to make the DBI or drivers even more efficient, I'm all ears.)

Why doesn't my CGI script work right?

Read the information in the references below. Please do not post CGI related questions to the dbi-users mailing list (or to me).

General problems and good ideas:

 Use the CGI::ErrorWrap module.
 Remember that many env vars won't be set for CGI scripts

How can I maintain a WWW connection to a database?

For information on the Apache httpd server and the mod_perl module see

A driver build fails because it can't find DBIXS.h

The installed location of the DBIXS.h file changed with 0.77 (it was being installed into the 'wrong' directory but that's where driver developers came to expect it to be). The first thing to do is check to see if you have the latest version of your driver. Driver authors will be releasing new versions which use the new location. If you have the latest then ask for a new release. You can edit the Makefile.PL file yourself. Change the part which reads "-I.../DBI" so it reads "-I.../auto/DBI" (where ... is a string of non-space characters).

Has the DBI and DBD::Foo been ported to NT / Win32?

The latest version of the DBI and, at least, the DBD::Oracle module will build - without changes - on NT/Win32 if your are using the standard Perl 5.004 and not the ActiveWare port.

Jeffrey Urlwin (or ) is helping me with the port (actually he's doing it and I'm integrating the changes :-).

What about ODBC?

A DBD::ODBC module is available.

Does the DBI have a year 2000 problem?

No. The DBI has no knowledge or understanding of dates at all.

Individual drivers (DBD::*) may have some date handling code but are unlikely to have year 2000 related problems within their code. However, your application code which uses the DBI and DBD drivers may have year 2000 related problems if it has not been designed and writtem well.

See also the ``Does Perl have a year 2000 problem?'' section of the Perl FAQ:


 Author:  Tim Bunce

Oracle - DBD::Oracle
 Author:  Tim Bunce

Ingres - DBD::Ingres
 Author:  Henrik Tougaard

DB2 - DBD::DB2
Empress - DBD::Empress
Informix - DBD::Informix
 Author:  Jonathan Leffler

Solid - DBD::Solid
 Author:  Thomas Wenrich

Postgres - DBD::Pg
 Author:  Edmund Mergl

Illustra - DBD::Illustra
 Author:  Peter Haworth

Fulcrum SearchServer - DBD::Fulcrum
 Author:  Davide Migliavacca


Apache::DBI by
To be used with the Apache daemon together with an embedded perl interpreter like mod_perl. Establishes a database connection which remains open for the lifetime of the http daemon. This way the CGI connect and disconnect for every database access becomes superfluous.

JDBC Server by Stuart 'Zen' Bishop
The server is written in Perl. The client classes that talk to it are of course in Java. Thus, a Java applet or application will be able to comunicate via the JDBC API with any database that has a DBI driver installed. The URL used is in the form jdbc:dbi://host.domain.etc:999/Driver/DBName. It seems to be very similar to some commercial products, such as jdbcKona.

Remote Proxy DBD support
  Carl Declerck <>
  Terry Greenlaw <>

Carl is developing a generic proxy object module which could form the basis of a DBD::Proxy driver in the future. Terry is doing something similar.

SQL Parser
        Hugo van der Sanden <>
        Stephen Zander <>

Based on the O'Reilly lex/yacc book examples and byacc.