Profiles and profile customization are introduced in "SQLJ Profiles (ISO Standard Code)". This appendix presents more technical detail and discusses customizer options and how to use customizers other than the default Oracle customizer.
There is also discussion of Oracle specialized customizers, particularly the SQLCheckerCustomizer
for semantics-checking profiles, and the AuditorInstaller
for installing "auditors" for debugging.
The following topics are covered:
Note:
If you use the default Oracle-specific code generation (-codegen=oracle
), the discussion in this appendix does not pertain to your application.SQLJ profiles contain information about your embedded SQL operations, with a separate profile being created for each connection context class that your application uses. Profiles are created during the SQLJ translator code generation phase and customized during the customization phase. Customization enables your application to use vendor-specific database features. Separating these vendor-specific operations into your profiles enables the rest of your generated code to remain generic.
Each profile contains a series of entries for the SQLJ statements that use the relevant connection context class, where each entry corresponds to one SQL operation in your application.
Profiles exist as serialized objects stored in resource files packaged with your application. Because of this, profiles can be loaded, read, and modified (added to or re-customized) at any time. When profiles are customized, information is only added, never removed. Multiple customizations can be made without losing preceding customizations, so that your application maintains the capability to run in multiple environments. This is known as binary portability.
For profiles to have binary portability, SQLJ industry-standard requirements have been met in the Oracle SQLJ implementation.
During code generation, the translator creates each profile as follows:
It creates a profile object as an instance of the sqlj.runtime.profile.Profile
class.
It inserts information about your embedded SQL operations into the profile object, for SQLJ statements that use the relevant connection context class.
It serializes the profile object into a Java resource file, referred to as a profile file, with a .ser
file name extension.
Note:
The Oracle SQLJ implementation provides an option to have the translator automatically convert these.ser
files to .class
files. The.ser
files are not supported by some browsers, and can be cumbersome when loading translated applications into the server. However, this prevents any further customization of the profile. For information, see "Conversion of .ser File to .class File (-ser2class)".As discussed in "Code Generation", profile file names for application Foo
are of the form:
Foo_SJProfilen.ser
SQLJ generates Foo_SJProfile0.ser
, Foo_SJProfile1.ser
, and so on, as needed, depending on how many connection context classes you use in your code. Or, if the -ser2class
option is enabled, then SQLJ generates Foo_SJProfile0.class
, Foo_SJProfile1.class
, and so on.
Each profile has a getConnectedProfile()
method that is called during SQLJ runtime. This method returns something equivalent to a JDBC Connection
object, but with added functionality. This is further discussed in "Functionality of a Customized Profile at Run Time".
Note:
Referring to a "profile object" indicates that the profile is in its original non-serialized state. Referring to a "profile file" indicates that the profile is in its serialized state in a.ser
file.Following is a sample SQLJ executable statement with the profile entry that would result. For simplicity, the profile entry is presented as plain text with irrelevant portions omitted.
Note that in the profile entry, the host variable is replaced by JDBC syntax (the question mark).
Presume the following declaration:
#sql iterator Iter (double sal, String ename);
And presume the following executable statements:
String empname = 'Smith'; Iter it; ... #sql it = { SELECT ename, sal FROM emp WHERE ename = :empname };
================================================================= ... #sql { SELECT ename, sal FROM emp WHERE ename = ? }; ... PREPARED_STATEMENT executed through EXECUTE_QUERY role is QUERY descriptor is null contains one parameter 1. mode: IN, java type: java.lang.String (java.lang.String), sql type: VARCHAR, name: ename, ... result set type is NAMED_RESULT result set name is Iter contains 2 result columns 1. mode: OUT, java type: double (double), sql type: DOUBLE, name: sal, ... 2. mode: OUT, java type: java.lang.String (java.lang.String), sql type: VARCHAR, name: ename, ... =================================================================
Note:
This profile entry is presented here as text for convenience only; profiles are not actually in text format. They can be printed as text, however, using the SQLJ-P-print
option, as discussed in "Overview of Customizer Harness Options".When using ISO SQLJ code, running the sqlj
script on a SQLJ source file includes an automatic customization process, where each profile created during the code generation phase is customized for use with your particular database. The default customizer is the Oracle customizer, oracle.sqlj.runtime.OraCustomizer
, which optimizes your profiles to use type extensions and performance enhancements specific to Oracle10i.
You can also run the sqlj
script to customize profiles created previously. On the SQLJ command line, you can specify .ser
files individually, JAR files containing .ser
files, or both.
Notes:
Whenever you use the default Oracle customizer during translation, your application will require the Oracle SQLJ run time and an Oracle JDBC driver when it runs, even if you do not use Oracle extensions in your code.
If an application has no customizations, or none suitable for the connection, then the generic SQLJ run time is used.
You can run SQLJ to process .sqlj
and .java
files (for translation, compilation, and customization) or to process .ser
and .jar
files (for customization only), but not both categories at once.
Regardless of whether you use the Oracle customizer or an alternative customizer, SQLJ uses a front-end customization utility known as the customizer harness in accomplishing your customizations.
When you run SQLJ, you can specify customization options for the customizer harness (for general customization settings that apply to any customizer you use) and for your customizer (for settings used by the particular customizer). In either case, you can specify these options either on the command line or in a properties file. This is discussed in "Customization Options and Choosing a Customizer".
A customizer is required to be a JavaBeans component adhering to the standard JavaBeans API to expose its properties, and must implement the sqlj.runtime.profile.util.ProfileCustomizer
interface, which specifies a customize()
method. For each profile to be customized, the customizer harness calls the customize()
method of the customizer object.
The Oracle customizer meets the preceding requirements and is defined in the oracle.sqlj.runtime.OraCustomizer
class.
The SQLJ customization process during translation consists of the following steps, as applicable, either during the customization stage of an end-to-end SQLJ run, or when you run SQLJ to customize existing profiles only:
SQLJ instantiates and invokes the customizer harness and passes it any general customization options you specified.
The customizer harness instantiates the customizer you are using and passes it any customizer-specific options you specified.
When you run SQLJ for customization only, specifying one or more JAR files on the command line, the customizer harness discovers and extracts the profile files within these JAR files.
The customizer harness deserializes each profile file into a profile object (.ser
files automatically created during an end-to-end SQLJ run, .ser
files specified on the command line for customization only, or .ser
files extracted from JAR files specified on the command line for customization only).
If the customizer you use requires a database connection, the customizer harness establishes that connection.
For each profile, the harness calls the customize()
method of the customizer object instantiated in step 2 (customizers used with SQLJ must have a customize()
method).
For each profile, the customize()
method typically creates and registers a profile customization within the profile. This depends on the intended functionality of the customizer, however. Some might have a specialized purpose that does not require a customization to be created and registered in this way.
The customizer harness reserializes each profile and puts it back into a .ser
file.
When you run SQLJ for customization only, specifying one or more JAR files on the command line, the customizer harness re-creates the JAR contents, inserting each customized .ser
file to replace the original corresponding uncustomized .ser
file.
Notes:
If an error occurs during customization of a profile, the original .ser
file is not replaced.
If an error occurs during customization of any profile in a JAR file, the original JAR file is not replaced.
SQLJ can run only one customizer at a time. If you want to accomplish multiple customizations on a single profile, you must run SQLJ multiple times. For the additional customizations, enter the profile name directly on the SQLJ command line.
When the harness calls the customize()
method to customize a profile, it passes in the profile object, a SQLJ connection context object (if you are using a customizer that requires a connection), and an error log object (which is used in logging error messages during the customization).
The same error log object is used for all customizations throughout a single running of SQLJ, but its use is transparent. The customizer harness reads messages written to the error log object and reports them in real-time to the standard output device (whatever SQLJ uses, typically your screen).
Recall that each profile has a set of entries, where each entry corresponds to a SQL operation. (These would be the SQL operations in your application that use instances of the connection context class associated with this profile.)
A customize()
method implements special processing on these entries. It could be as simple as checking each entry to verify its syntax, or it could be more complicated, such as creating new entries that are equivalent to the original entries but are modified to use features of your particular database.
Notes:
Any customize()
processing of profile entries does not alter the original entries.
Customizing your profiles for use in a particular environment does not prevent your application from running in a different environment. You can customize a profile multiple times for use in multiple environments, and these customizations will not interfere with each other.
The customizer harness outputs error and status messages in much the same way as the SQLJ translator, outputting them to the same output device. None of the warnings regarding customization are suppressible, however.
Error messages reported by the customizer harness fall into four categories:
Unrecognized or illegal option
Connection instantiation error
Profile instantiation error
Customizer instantiation error
Status messages reported by the customizer harness during customization allow you to determine whether a profile was successfully customized. They fall into three categories:
Profile modification status
JAR file modification status
Name of backup file created (if the customizer harness backup
option is enabled)
Additional customizer-specific errors and warnings might be reported by the customize()
method of the particular customizer.
During customization, the profile customizer writes messages to its error log, and the customizer harness reads the log contents in real-time and outputs these messages to the SQLJ output device, along with any other harness output. You never have to access error log contents directly.
A customized profile is a static member of the connection context class with which it is associated. For each SQLJ statement in your application, the SQLJ run time determines the connection context class and instance associated with that statement, then uses the customized profile of the connection context class, together with the underlying JDBC connection of the particular connection context instance, to create a connected profile. This connected profile is the vehicle that the SQLJ run time uses in applying vendor-specific features to the execution of your SQLJ application.
This section discusses options for profile customization, which fall into three categories:
Options you specify to the customizer harness, which apply to whatever customizer you use
This includes general options, connection options, and options that invoke specialized customizers.
Customizer-specific options you specify to your customizer through the customizer harness
SQLJ options, which determine basic aspects of customization, such as whether to customize at all and which customizer to use
All categories of options are specified through the SQLJ command line or properties files.
The following topics are included in this section:
To choose a customizer other than the default Oracle customizer, you can use either the customizer harness customizer
option (discussed in "Overview of Customizer Harness Options") or the SQLJ -default-customizer
option (discussed in "SQLJ Translator Options for Profile Customization").
The customizer harness provided with the Oracle SQLJ implementation offers a number of options that are not specific to a particular customizer. The harness uses these options in its front-end coordination of the customization process.
Customizer harness option settings on the SQLJ command line have the following syntax:
-P-option=value
Alternatively, in a SQLJ properties file:
profile.option=value
Enable boolean options (flags) either with:
-P-option
or:
-P-option=true
Boolean options are disabled by default, but you can explicitly disable them with:
-P-option=false
This option syntax is also discussed in "Options to Pass to the Profile Customizer (-P)" and "Properties File Syntax".
The customizer harness supports the following general options:
backup
: Save a backup copy of the profile before customizing it.
context
: Limit customizations to profiles associated with the listed connection context classes.
customizer
: Specify the customizer to use.
digests
: Specify digests for JAR file manifests (relevant only if specifying JAR files to customize).
help
: Display customizer options (specified only in SQLJ command line).
verbose
: Display status messages during customization.
The customizer harness supports the following options for customizer database connections. Currently, these are used by the Oracle customizer if you enable its optcols
option for column definitions (for performance optimization). In addition, they are used by the SQLCheckerCustomizer
if you use this specialized customizer to perform online semantics-checking on profiles.
user
: Specify the user name for the connection used in this customization.
password
: Specify the password for the connection used in this customization.
url
: Specify the URL for the connection used in this customization.
driver
: Specify the JDBC driver for the connection used in this customization.
For information about the Oracle customizer optcols
flag, see "Oracle Customizer Column Definition Option (optcols)". For information about the SQLCheckerCustomizer
, see "SQLCheckerCustomizer for Profile Semantics-Checking".
The following commands function as customizer harness options, but are implemented through specialized customizers provided with the Oracle SQLJ implementation.
debug
: Insert debugging information into the specified profiles, to be output at run time. This is a shortcut to invoke the Oracle SQLJ AuditorInstaller
, which is described in "AuditorInstaller Customizer for Debugging".
print
: Output the contents of the specified profiles, in text format.
verify
: Perform semantics-checking on a profile that was produced during a previous execution of the SQLJ translator (equivalent to semantics-checking performed on source code during translation). This is a shortcut to invoke the Oracle SQLJ SQLCheckerCustomizer
, which is described in "SQLCheckerCustomizer for Profile Semantics-Checking".
This section describes general options supported by the customizer harness.
Use the backup
flag to instruct the harness to save a backup copy of each .jar
file and standalone .ser
file before replacing the original. (Separate backups of .ser
files that are within .jar
files are not necessary.)
Backup file names are given the extension .bak
n
, where n
indicates digits used as necessary where there are similarly named files. For each backup file created, an informational message is issued.
If an error occurs during customization of a standalone .ser
file, then the original .ser
file is not replaced and no backup is created. Similarly, if an error occurs during customization of any .ser
file within a JAR file, then the original JAR file is not replaced and no backup is created.
Command-line syntax -P-backup<=true|false>
Command-line example -P-backup
Properties file syntax profile.backup<=true|false>
Default value false
Use the context
option to limit customizations to profiles that correspond to the specified connection context classes. Fully qualify the class names and use a comma-delimited list to specify multiple classes. For example:
-P-context=sqlj.runtime.ref.DefaultContext,foo.bar.MyCtxtClass
There must be no space on either side of the comma.
If this option is not specified, then all profiles are customized, regardless of their associated connection context classes.
Command-line syntax
-P-context=ctx_class1<,ctx_class2,...>
Command-line example
-P-context=foo.bar.MyCtxtClass
Properties file syntax
profile.context=ctx_class1<,ctx_class2,...>
Properties file example
profile.context=foo.bar.MyCtxtClass
Use the customizer
option to specify which customizer to use. Fully qualify the class name, such as in the following example:
-P-customizer=oracle.sqlj.runtime.util.OraCustomizer
If you do not set this option, then SQLJ will use the customizer specified in the SQLJ -default-customizer
option. Unless set otherwise, this is the following:
oracle.sqlj.runtime.util.OraCustomizer
Command-line syntax
-P-customizer=customizer_class
Command-line example
-P-customizer=a.b.c.MyCustomizer
Properties file syntax
profile.customizer=customizer_class
Properties file example
profile.customizer=a.b.c.MyCustomizer
Default value
None
When a JAR file is produced, the JAR utility can optionally include one or more digests for each entry, based on one or more specified algorithms, so that the integrity of the JAR file entries can later be verified. Digests are similar conceptually to checksums, for readers familiar with those.
If you are customizing profiles in a JAR file and want the JAR utility to add new digests (or update existing digests) when the JAR file is updated, use the digests
option to specify a comma-delimited list of one or more algorithms. These are the algorithms that the JAR utility will use in creating the digests for each entry. The JAR utility produces one digest for each algorithm for each JAR file entry in the JAR manifest file. Specify algorithms as follows:
-P-digests=SHA,MD5
There must be no space on either side of the comma.
In this example, there will be two digests for each entry in the JAR manifest file: an SHA
digest and an MD5
digest.
For information about JAR files and the JAR utility, see one of the following Web sites:
http://www.javasoft.com/products/jdk/1.2/docs/guide/jar/index.html
or:
http://java.sun.com/j2se/1.3/docs/guide/jar/index.html
Use the help
option to display the option lists of the customizer harness and the default customizer or a specified customizer. For the harness and Oracle customizer, this includes a brief description and the current setting of each option.
Display the option lists for the harness and default customizer as follows (where the default customizer is the Oracle customizer or whatever you have specified in the SQLJ -default-customizer
option):
-P-help
Use the help
option in conjunction with the customizer
option to display the option list of a particular customizer, as follows:
-P-help -P-customizer=sqlj.runtime.profile.util.AuditorInstaller
Notes:
You can use the -P-help
option on the SQLJ command line only, not in a SQLJ properties file.
No customizations are performed if the -P-help
flag is enabled, even if you specify profiles to customize on the command line.
This section describes connection options supported by the customizer harness. These are used as follows:
The Oracle customizer uses database connections only for column definitions. If you do not enable the Oracle customizer optcols
option, then there is no need to set the customizer harness user
, password
, url
, and driver
options.
The SQLCheckerCustomizer
, a specialized customizer that performs semantics-checking on profiles, uses the customizer harness user
, password
, url
, and driver
settings for online checking.
Use -P-verify
on the SQLJ command line to invoke this customizer.
Note:
Do not confuse the customizer harnessuser
, password
, url
, and driver
options with the translator options of the same names, which are for semantics-checking during the translation step. However, the translator settings are passed to the customizer for convenience, in case customization is to use the same connection as translation. Override these initial settings through the customizer harness options if you wish.Set the user
option to specify a database schema if your customizer uses database connections.
In addition to specifying the schema, you can optionally specify the password, URL, or both in your user
option setting. The password is preceded by a forward-slash (/), and the URL is preceded by an "at" sign (@), as in the following examples:
-P-user=scott/tiger -P-user=scott@jdbc:oracle:oci:@ -P-user=scott/tiger@jdbc:oracle:oci:@
Note:
When you use column definitions (optcols
option), the user setting for the SQLJ translator is forwarded to the profile customizer as well, but you can use the customizer user
option to override the translator setting.Use the password
option if your customizer uses database connections.
The password can also be set with the user
option, as described in "Customization User Option (user)".
Note:
When you use column definitions (optcols
option), the password setting for the SQLJ translator is forwarded to the profile customizer as well, but you can use the customizer password
option to override the translator setting.Use the url
option if your customizer uses database connections.
The URL can also be set with the user
option, as described in "Customization User Option (user)".
Note:
When you use column definitions (optcols
option), the URL setting for the SQLJ translator is forwarded to the profile customizer as well, but you can use the customizer url
option to override the translator setting.The customizer harness supports the following options that invoke specialized customizers:
debug
: This invokes the AuditorInstaller
customizer, used in debugging.
print
: This invokes a customizer that prints a text version of a profile.
verify
: This invokes the SQLCheckerCustomizer
customizer, which performs semantics-checking on a profile.
Important:
Because each of these options invokes a customizer, and only one customizer can run in a single execution of SQLJ, you cannot perform any other customization when you use any of these options.You also cannot use more than one of print
, debug
, or verify
simultaneously.
The debug
option runs a specialized customizer, called the AuditorInstaller
, that inserts debugging statements into profiles. Use this option in conjunction with a SQLJ command line file list to insert debugging statements into the specified profiles. These profiles must already be customized from a previous SQLJ run.
For detailed information about this customizer, including additional options that it supports, see "AuditorInstaller Customizer for Debugging".
The debugging statements will execute during SQLJ run time (when someone runs your application), displaying a trace of method calls and values returned.
Following are examples of how to specify the debug
option:
% sqlj -P-debug Foo_SJProfile0.ser Bar_SJProfile0.ser % sqlj -P-debug *.ser
The print
option runs a specialized customizer that prints profiles in text format. Use this option in conjunction with a SQLJ command line file list to output the contents of one or more specified profiles. The output goes to the standard SQLJ output device, typically the user screen.
Following are examples of how to specify the print
option:
% sqlj -P-print Foo_SJProfile0.ser Bar_SJProfile0.ser % sqlj -P-print *.ser
For sample output, see "Sample Profile Entry".
The verify
option runs a specialized customizer, called the SQLCheckerCustomizer
, that performs semantics-checking on a profile. This is equivalent to the semantics-checking that is performed on source code during translation. The profile will have been created during a previous execution of the SQLJ translator.
This option is useful for checking semantics against the run time database, after deployment, and after the source code may no longer be available.
For detailed information about this customizer, including additional options that it supports, see "SQLCheckerCustomizer for Profile Semantics-Checking".
Note:
For online semantics-checking of the profile, you must also use the customizer harnessuser
, password
, and url
options.Following are examples of how to specify the verify
option. Both of these examples use the SQLCheckerCustomizer
default semantics-checker, which employs online checking through the specified database connection. (The first is a single wraparound command.)
% sqlj -P-verify -P-user=scott -P-password=tiger -P-url=jdbc:oracle:oci:@ Foo_SJProfile0.ser Bar_SJProfile0.ser % sqlj -P-verify -P-user=scott -P-password=tiger -P-url=jdbc:oracle:oci:@ *.ser
You can set customizer-specific options, such as options for the Oracle customizer, on the SQLJ command line or in a SQLJ properties file. The syntax is similar to that for setting customizer harness options.
Set a customizer option on the SQLJ command line by preceding it with:
-P-C
Alternatively, you can set it in a SQLJ properties file by preceding it with:
profile.C
This option syntax is also discussed in "Options to Pass to the Profile Customizer (-P)" and "Properties File Syntax".
The remainder of this section discusses features of the Oracle customizer, which supports several options. Most of these options are boolean and are enabled as follows:
-P-Coption
or:
-P-Coption=true
Boolean options are disabled by default, but you can explicitly disable them with:
-P-Coption=false
Numeric or string options are set similarly:
-P-Coption=value
This section describes options that are specific to the Oracle customizer, beginning with an overview of the options supported.
The Oracle customizer implements the following options:
compat
: Display version-compatibility information.
force
: Instruct the customizer to customize even if a valid customization already exists.
optcols
: Enable iterator column type and size definitions to optimize performance.
optparams
: Enable parameter size definitions to optimize JDBC resource allocation (used in conjunction with optparamdefaults
).
optparamdefaults
: Set parameter size defaults for particular data types (used in conjunction with optparams
).
fixedchar
: Enable CHAR
comparisons with blank padding for WHERE
clauses.
showSQL
: Display SQL statement transformations.
stmtcache
: Set the statement cache size (the number of statements that can be cached for each connection during run time) for performance optimization, or set it to zero to disable statement caching.
summary
: Display a summary of Oracle features used in your application.
Any output displayed by these options is written to the standard output device, wherever SQLJ writes its other messages.
Use the compat
flag to instruct the Oracle customizer to display information about compatibility of your application with different versions of the Oracle database and Oracle JDBC drivers. This can be accomplished either during a full SQLJ translation run or on profiles previously created.
For example, to see compatibility output when translating and customizing the application MyApp
:
% sqlj <...SQLJ options...> -P-Ccompat MyApp.sqlj
In this example, the MyApp
profiles will be created, customized, and checked for compatibility in a single running of SQLJ.
To see compatibility output for MyApp
profiles previously created:
% sqlj <...SQLJ options...> -P-Ccompat MyApp_SJProfile*.ser
In this example, the MyApp
profiles were created (and possibly customized) in a previous running of SQLJ and will be customized (if needed) and checked for compatibility in the above running of SQLJ.
Following are two output samples from a -P-Ccompat
setting when using the default Oracle customizer. The first example indicates that the application can be used with all Oracle JDBC driver versions:
MyApp_SJProfile0.ser: Info: compatible with all Oracle JDBC drivers
This second example indicates that the application can be used only with the JDBC implementation from an Oracle 8.1.x or later release:
MyApp_SJProfile0.ser: Info: compatible with Oracle 8.1 or later JDBC driver
Note:
If customization does not take place because a valid previous customization is detected, thecompat
option reports compatibility regardless.Use the force
flag to instruct the Oracle customizer to force the customization of a given profile (specified on the command line) even if a valid customization already exists in that profile. For example:
% sqlj -P-Cforce MyApp_SJProfile*.ser
This will customize all the MyApp
profiles, regardless of whether they have already been customized. Otherwise, by default, the Oracle customizer will not reinstall over a previously existing customization unless the previous one had been installed with an older version of the customizer.
Use the optcols
flag to instruct the Oracle customizer to determine types and sizes of iterator or result set columns and add this information to the profile. This enables the SQLJ run time to automatically register the columns with the Oracle JDBC driver when your application runs, saving round trips to Oracle depending on the particular driver implementation. Specifically, this is effective for the Thin driver and positional iterators.
For an overview of column definitions, see "Column Definitions".
An error will be generated if you enable the Oracle customizer optcols
option without setting the user name, password, and URL for a database connection. You can accomplish this through the translator -user
, -password
, and -url
options, which are forwarded to the customizer during ISO standard code generation, or directly through the customizer user
, password
, and url
options.
The customizer does not have to connect to the same schema or even the same database that your application will connect to at run time, but the relevant columns will have to be in the same order and of identical types and sizes to avoid run time errors.
For information about the customizer harness connection options, see the user
, password
, url
, and driver
sections under "Overview of Customizer Harness Options".
Note:
You can use the SQLJ translator-optcols
option instead. This sets the customizer option automatically. (And for Oracle-specific code generation, which uses no profiles, you must use the translator option instead.) See "Column Definitions (-optcols)".
That section also has some additional conceptual information.
You can enable or disable the customizer optcols
flag on the SQLJ command line or in a properties file.
Enable it on the command line as follows:
-P-Coptcols
or:
-P-Coptcols=true
This flag is disabled by default, but you can also disable it explicitly. Disable it on the command line as follows:
-P-Coptcols=false
Column definitions require the customizer to make a database connection to examine columns of tables being queried, so the customizer harness user
, password
, and url
options must be set appropriately (as well as the customizer harness driver
option if you are not using the default OracleDriver
class). For example:
% sqlj <...SQLJ options...> -P-user=scott/tiger@jdbc:oracle:oci:@ -P-Coptcols MyApp.sqlj
Note that as with the SQLJ translator, you can optionally set the password and URL in the user
option instead of in the password
and url
options.
Alternatively, you can insert column definitions into a previously existing profile. In this case you must also use the Oracle customizer force
option to force a recustomization:
% sqlj -P-user=scott/tiger@jdbc:oracle:oci:@ -P-Cforce -P-Coptcols MyApp_SJProfile*.ser
You also can insert column definitions into previously existing profiles in a JAR file:
% sqlj -P-user=scott/tiger@jdbc:oracle:oci:@ -P-Cforce -P-Coptcols MyAppProfiles.jar
When you run the Oracle customizer with its optcols
flag enabled, either during translation and creation of a new profile or during customization of an existing profile, you can also enable the customizer harness verbose
flag. This will instruct the Oracle customizer to display information about what iterators and result sets are being processed and what their column type and size definitions are. For example:
% sqlj -P-user=scott/tiger@jdbc:oracle:oci:@ -P-verbose -P-Cforce -P-Coptcols MyApp_SJProfile*.ser
For general information about the verbose
flag, see that section under "Overview of Customizer Harness Options".
You can execute the Oracle customizer with its summary
flag enabled on an existing profile to determine if column definitions have been added to that profile:
% sqlj -P-Csummary MyApp_SJProfile*.ser
For general information about the summary
flag, see that section under "Overview of Customizer-Specific Options".
Use the optparams
flag to enable parameter size definitions. If this flag is enabled, SQLJ will register your input and output parameters (host variables) to optimize JDBC resource allocations according to sizes you specify.
For an overview of parameter size definitions and a discussion of source code hints, see "Parameter Size Definitions".
Note:
You can use the SQLJ translator-optparams
option instead. This sets the customizer option automatically. (And for Oracle-specific code generation, which uses no profiles, you must use the translator option instead.) See "Parameter Definitions (-optparams)".
That section also has some additional conceptual information.
You can enable or disable the optparams
flag on the command line or in a SQLJ properties file.
Enable it on the command line as follows:
-P-Coptparams
or:
-P-Coptparams=true
This flag is disabled by default, but you can also disable it explicitly. Disable it on the command line as follows:
-P-Coptparams=false
Note:
Unlike theoptcols
option, the optparams
option does not require a database connection by the customizer, because you are providing the size specifications yourself.Following is a command-line example (omitting a setting for the optparamdefaults
option, which is discussed in the next section):
% sqlj <...SQLJ options...> -P-Coptparams -P-Coptparamdefaults=defaults_string MyApp.sqlj
Alternatively, to enable parameter size definitions for a previously existing profile:
% sqlj -P-Coptparams -P-Coptparamdefaults=defaults_string MyApp_SJProfile*.ser
You can also use previously existing profiles in a JAR file:
% sqlj -P-Coptparams -P-Coptparamdefaults=defaults_string MyAppProfiles.jar
If you enable the optparams
option to set parameter sizes, use the optparamdefaults
option as desired to set default sizes for specified data types. If optparams
is not enabled, then any optparamdefaults
setting is ignored.
For an overview of parameter size definitions and a discussion of source code hints, see "Parameter Size Definitions".
Note:
You can use the SQLJ translator-optparamdefaults
option instead. This sets the customizer option automatically. (And for Oracle-specific code generation, which uses no profiles, you must use the translator option instead.) See "Parameter Default Size (-optparamdefaults)".
That section also has important additional conceptual and syntax information. Functionality of the two options is equivalent.
You can set the optparamdefaults
flag on the command line or in a SQLJ properties file.
Set it on the command line as follows:
-P-Coptparamdefaults=datatype1(size1),datatype2(size2),...
Following is a command-line example, including the optparams
setting as well:
% sqlj <..SQLJ options..> -P-Coptparams -P-Coptparamdefaults=CHAR_TYPE(50),RAW_TYPE(500),CHAR(10) MyApp.sqlj
The syntax is explained in "Parameter Default Size (-optparamdefaults)".
Alternatively, you can specify parameter size defaults for a previously existing profile, in which case you must also use the Oracle customizer force
option to force a recustomization:
% sqlj -P-Cforce -P-Coptparams -P-Coptparamdefaults=CHAR_TYPE(50),RAW_TYPE(500),CHAR(10) MyApp_SJProfile*.ser
You also can specify parameter size defaults for previously existing profiles in a JAR file:
% sqlj -P-Cforce -P-Coptparams -P-Coptparamdefaults=CHAR_TYPE(50),RAW_TYPE(500),CHAR(10) MyAppProfiles.jar
Note:
If at run time the actual size exceeds the registered size of any parameter, run time errors will occur.Set this flag to true
to account for blank padding in CHAR
database columns when binding character strings for WHERE
clause comparisons. This way, for example, "mystring" would compare positively against "mystring ".
Here is an example of Oracle customizer fixedchar
usage:
% sqlj -P-Cfixedchar MyProgram.sqlj AnotherProg.java ...
Notes:
You can use the SQLJ translator -fixedchar
option instead. This sets the customizer option automatically. (And for Oracle-specific code generation, which uses no profiles, you must use the translator option instead.) See "CHAR Comparisons with Blank Padding (-fixedchar)". That section also has some additional conceptual information.
If you also enable the Oracle customizer summary
flag, the number of usages of the Oracle setFixedCHAR()
API (used behind the scenes for fixedchar
functionality) will be displayed. See "Oracle Customizer Summary Option (summary)" for an example.
Use the showSQL
flag to display any SQL statement transformations performed by the Oracle customizer. Such transformations are necessary in cases where SQLJ supports syntax that Oracle10i does not.
To show SQL transformations when translating and customizing the application MyApp
:
% sqlj <...SQLJ options...> -P-CshowSQL MyApp.sqlj
In this example, the MyApp
profiles will be created and customized and their SQL transformations displayed in a single running of SQLJ.
To show SQL transformations when customizing MyApp
profiles previously created:
% sqlj <...SQLJ options...> -P-CshowSQL MyApp_SJProfile*.ser
In this example, the MyApp
profiles were created (and possibly customized) in a previous running of SQLJ and will be customized (if needed) and have their SQL transformations displayed in the above running of SQLJ.
The showSQL
output might include an entry such as this:
MyApp.sqlj:14: Info: <<<NEW SQL>>> #sql {BEGIN ? := VALUES(tkjsSET_f1); END}; in file MyApp, line 14, we had:
#sql {set :v1= VALUES(tkjsSET_f1) };
SQLJ supports the SET
statement, but Oracle10i does not. During customization, the Oracle customizer replaces the SET
statement with an equivalent PL/SQL block.
Note:
If customization does not take place because a valid previous customization is detected, theshowSQL
option shows SQL transformations regardless.Use the Oracle customizer stmtcache
option to set the statement cache size—the number of statements that can be cached for each database connection as your application runs—or to disable statement caching.
The default statement cache size is 5. For an overview of statement caching, see "Statement Caching".
Important:
With the default Oracle-specific code generation (-codegen=oracle
), SQLJ does not produce profiles and skips the customization step. In this case, use connection context methods to control SQLJ statement caching. See "Connection Context Methods for Statement Caching (Oracle-Specific Code)".You can set the statement cache size on the command line or in a properties file.
To use the command line to set the statement cache size to 15 (for example) for the application MyApp
:
% sqlj <...SQLJ options...> -P-Cstmtcache=15 MyApp.sqlj
To disable statement caching, set the cache size to 0:
% sqlj <...SQLJ options...> -P-Cstmtcache=0 MyApp.sqlj
You also can alter the statement cache size in an existing profile without re-translating the application, but you must also use the Oracle customizer force
option to force a recustomization, as follows:
% sqlj -P-Cforce -P-Cstmtcache=15 MyApp_SJProfile0.ser
If you have multiple profiles, you can set their statement cache sizes individually by running SQLJ separately for each profile, after you have translated your application:
% sqlj -P-Cforce -P-Cstmtcache=10 MyApp_SJProfile0.ser % sqlj -P-Cforce -P-Cstmtcache=15 MyApp_SJProfile1.ser % sqlj -P-Cforce -P-Cstmtcache=0 MyApp_SJProfile2.ser
Of course, you must determine which profile corresponds to each of your connection context classes. This is determined as follows: profile 0 will correspond to the connection context class used for the first executable statement in your application; profile 1 will correspond to the connection context class used for the first executable statement that does not use the first connection context class, and so on. You can verify the correlation by using the customizer harness print
option to examine each profile.
Use the summary
flag to instruct the Oracle customizer to display a summary of Oracle features used in an application being translated, or in specified profile files. This is useful in identifying features that would prevent portability to other platforms and can be accomplished either during a full SQLJ translation run or on profiles previously created.
To see summary output when translating and customizing the application MyApp
:
% sqlj <...SQLJ options...> -P-Csummary MyApp.sqlj
In this example, the MyApp
profiles will be created, customized, and summarized in a single running of SQLJ.
To see summary output for MyApp
profiles previously created:
% sqlj <...SQLJ options...> -P-Csummary MyApp_SJProfile*.ser
In this example, the MyApp
profiles were created (and possibly customized) in a previous running of SQLJ and will be customized (if needed) and summarized in the above running of SQLJ.
Following are two samples resulting from a -P-Csummary
setting when using the default Oracle customizer. The first example indicates no Oracle features are used:
MyApp_SJProfile0.ser: Info: Oracle features used: MyApp_SJProfile0.ser: Info: * none
This second example indicates that Oracle features are used—several Oracle extended data types from the oracle.sql
package—and lists them:
MyApp_SJProfile0.ser: Info: Oracle features used: MyApp_SJProfile0.ser: Info: * oracle.sql.NUMBER: 2 MyApp_SJProfile0.ser: Info: * oracle.sql.DATE: 2 MyApp_SJProfile0.ser: Info: * oracle.sql.CHAR: 2 MyApp_SJProfile0.ser: Info: * oracle.sql.RAW: 2
The following example prints out the number of usages of the Oracle setFixedCHAR()
API (enabled through the Oracle customizer fixedchar
option, to account for blank padding when binding a string into a WHERE
clause for comparison against CHAR
data):
% sqlj -P-Cfixedchar -P-Csummary -P-Cforce *.ser FC_SJProfile0.ser: Info: re-installing Oracle customization FC_SJProfile0.ser: Info: Oracle features used: FC_SJProfile0.ser: Info: * setFixedCHAR(): 4
Note:
If customization does not take place because a valid previous customization is detected, thesummary
option produces a summary regardless.The Oracle SQLJ implementation provides additional, specialized customizers described later in this chapter. These customizers also have command-line options:
SQLCheckerCustomizer
(for profile semantics-checking): See "SQLCheckerCustomizer for Profile Semantics-Checking" for general information, and "SQLCheckerCustomizer Options" for information about its options.
AuditorInstaller
(for debugging): See "AuditorInstaller Customizer for Debugging" for general information, and "AuditorInstaller Options" for information about its options.
The following SQLJ translator options relate to profile customization and are described elsewhere in this manual:
-default-customizer
: Specify the default profile customizer to use if none is specified in the customizer harness -customizer
option.
-profile
: Specify whether to customize during this running of SQLJ.
As discussed previously, you can specify a JAR file on the SQLJ command line in order to customize any profiles that the JAR file contains.
Notes:
Remember that you can specify .sqlj
or .java
files or both on the SQLJ command line for normal SQLJ processing, or you can specify .ser
or .jar
files or both on the command line for customization only, but not both categories.
It is permissible for the .jar
file to contain files that are not profiles. Any file whose manifest entry indicates that the file is not a profile will be ignored during customization.
The .jar
file is used as the class-loading context for each profile it contains. If a profile contains a reference to a class contained within the .jar
file, then that class is loaded from the .jar
file. If a profile contains a reference to a class not in the .jar
file, then the system class loader will find and load the class according to your classpath, as usual.
There are requirements for the manifest entry of each profile.
Create a plain text file with two lines for each profile that will be included in the JAR file. One line starts with "Name:", followed by the path or package and name. The other line is the following:
SQLJProfile: TRUE
The two lines must be consecutive (no blank line in between), and there must be a blank line preceding line-pairs for additional profiles.
Use the JAR utility -m
option to input this file.
For example, presume your MyApp
application (in the directory foo/bar
) has three profiles, and you will be creating a JAR file that will include these profiles. Complete the following steps:
Create a text file with the following eight lines (including the blank lines used as separators):
Name: foo/bar/MyApp_SJProfile0.ser SQLJProfile: TRUE Name: foo/bar/MyApp_SJProfile1.ser SQLJProfile: TRUE Name: foo/bar/MyApp_SJProfile2.ser SQLJProfile: TRUE
Presume you call this file MyAppJarEntries.txt
.
When you run jar
to create the JAR file, use the -m
option to input your text file as follows (presume you want to call the JAR file myjarfile.jar
):
% jar -cvfm myjarfile.jar MyAppJarEntries.txt foo/bar/MyApp_SJProfile*.ser foo/bar/*.class
As the JAR utility constructs the manifest during creation of the JAR file, it reads your text file and inserts the SQLJProfile: TRUE
line into the manifest entry of each profile. It accomplishes this by matching the names in the manifest with the names you specify in your text file.
When you specify a JAR file on the SQLJ command line, each profile in the JAR file is deserialized and customized.
A JAR file is successfully customized only if all the profiles it contains are successfully customized. After a successful customization, each profile has been reserialized into a .ser
file, the JAR file has been modified to replace the original .ser
files with the customized .ser
files, and the JAR file manifest has been updated to indicate the new entries.
If any error is encountered in the customization of any profile in a JAR file, then the JAR file customization has failed, and the original JAR file is left completely unchanged.
Note:
If you use signature files for authentication, the signature files that appeared in the original JAR file will appear unchanged in the updated JAR file. You are responsible for re-signing the new JAR file if the profiles require signing.Oracle provides a special customizer, SQLCheckerCustomizer
, that will perform semantics-checking on a profile that was produced during previous execution of the translator. This semantics-checking is similar to what is normally performed during translation of the source code.
This is particularly valuable when the database to be used at run time is not available for semantics-checking during translation. In these circumstances, you can use SQLCheckerCustomizer
after deployment, against the run time database, typically in a scenario where the source code is no longer available.
You can specify the checker to use. If you accept the default OracleChecker
front end, SQLCheckerCustomizer
will perform online semantics-checking using an appropriate online checker.
Note:
For online semantics-checking of the profile, you must also specify connection parameters using the customizer harness connection options.Following are examples of how to specify the Oracle customizer harness verify
option to run SQLCheckerCustomizer
in its default mode. Because it defaults to an online checker, you typically must provide connection parameters through the customizer harness user
, password
, and url
options. (The first example is a single wraparound command line.)
% sqlj -P-verify -P-user=scott -P-password=tiger -P-url=jdbc:oracle:oci:@ Foo_SJProfile0.ser Bar_SJProfile0.ser % sqlj -P-verify -P-user=scott -P-password=tiger -P-url=jdbc:oracle:oci:@ *.ser
The verify
option results in the customizer harness instantiating and invoking the following class:
sqlj.runtime.profile.util.SQLCheckerCustomizer
This class coordinates semantics-checking of the SQL operations in the profile. You can specify a semantics-checker or accept the default OracleChecker
semantics-checker front end.
The -P-verify
option is equivalent to the following:
-P-customizer=sqlj.runtime.profile.util.SQLCheckerCustomizer
This overrides the customizer specified in the SQLJ -default-customizer
option.
Notes:
As with any Oracle customizer, help output and an option list will be provided if you specify -P-verify
together with -P-help
on the SQLJ command line.
It is important to realize that because the verify
option invokes a customizer, and only one customizer can run in any single running of SQLJ, you cannot do any other customization when you use this option.
You also cannot use more than one of -P-print
, -P-debug
, and -P-verify
simultaneously, because each of these invokes a specialized customizer.
Like any customizer, SQLCheckerCustomizer
has its own options, which can be set using the -P-C
prefix on the SQLJ command line or the profile.C
prefix in a SQLJ properties file.
SQLCheckerCustomizer
supports the following options:
checker
: Specify the semantics-checker to use. The default is the OracleChecker
front end, as for checking during SQLJ translation.
warn
: Specify the categories of warnings and messages to display during semantics-checking of the profile. This is equivalent to the SQLJ -warn
flag for warning categories during translation-time semantics-checking, supports the same settings, and uses the same defaults. See "Translator Warnings (-warn)".
The checker
option enables you to specify the semantics-checker to use in checking the SQL operations in a profile.
This defaults to the Oracle semantics-checker front end, oracle.sqlj.checker.OracleChecker
, which for SQLCheckerCustomizer
chooses an appropriate online checker for your environment. For more information about OracleChecker
, see "Semantics-Checkers and the OracleChecker Front End (default checker)".
Following is a full command-line example, showing how to use the SQLCheckerCustomizer
checker
option, in conjunction with the customizer harness verify
option and connection options.
% sqlj -P-verify -P-user=scott -P-password=tiger -P-url=jdbc:oracle:oci:@ -P-Cchecker=abc.def.MyChecker *.ser
(This is a single wraparound command line.)
The warn
option is equivalent to the SQLJ translator -warn
option, allowing you to choose the categories of warnings and messages to be displayed as semantics-checking is performed on a profile.
For a complete description of the functionality and possible settings of these options, see "Translator Warnings (-warn)".
This defaults to the all,noverbose,noportable
settings, resulting in all warning categories except verbose
and portable
being enabled. You will receive any warnings regarding inheritance hierarchy requirements, data precision, conversion loss for nullable data, and strict matching for named iterators. These are the same defaults as for warnings during SQLJ translation.
Following is a full command-line example showing how to use the SQLCheckerCustomizer
warn
option, in conjunction with the customizer harness verify
option and connection options. This would result in only portability warnings being displayed.
% sqlj -P-verify -P-user=scott -P-password=tiger -P-url=jdbc:oracle:oci:@ -P-Cwarn=none,portable *.ser
(This is a single wraparound command line.)
For ISO code generation, SQLJ provides a special customizer, AuditorInstaller
. This customizer will insert sets of debugging statements, known as auditors, into profiles specified on the SQLJ command line. These profiles must already exist from previous customization.
The debugging statements will execute during SQLJ run time (when someone runs your application), displaying a trace of method calls and values returned.
Use the customizer harness debug
option, preceded by -P-
as with any general customization option, to insert the debugging statements. (Syntax for this option is discussed in "Invoking AuditorInstaller with the Customizer Harness debug Option".)
When an application is customized, the Oracle customizer implements profiles in layers of code (typically less than five) for different levels of run time functionality. The deepest layer uses straight Oracle JDBC calls and implements any of your SQLJ statements that can be executed through JDBC functionality. Each higher layer is a specialized layer for some category of SQLJ functionality that is not supported by JDBC and so must be handled specially by the SQLJ run time. For example, a layer for iterator conversion statements (CAST
) is used to convert JDBC result sets to SQLJ iterators. Another layer is used for assignment statements (SET
).
At run time, each SQLJ executable statement is first passed to the shallowest layer and then passed, layer-by-layer, until it reaches the layer that can process it (usually the deepest layer, which executes all JDBC calls).
You can install debugging statements at only one layer during a single execution of AuditorInstaller
. Each set of debugging statements installed at a particular layer of code is referred to as an individual auditor. During run time, an auditor is activated whenever a call is passed to the layer at which the auditor is installed.
Any one of the specialized code layers above the JDBC layer is usually of no particular interest during debugging, so it is typical to install an auditor at either the deepest layer or the shallowest layer. If you install an auditor at the shallowest layer, its run time debugging output will be a trace of method calls resulting from all your SQLJ executable statements. If you install an auditor at the deepest layer, its run time output will be a trace of method calls from all your SQLJ executable statements that result in JDBC calls.
Use multiple executions of AuditorInstaller
to install auditors at different levels. You might want to do that to install auditors at both the deepest layer and the shallowest layer, for example.
See "AuditorInstaller Depth Option (depth)" for information about how to specify the layer at which to install an auditor.
Following are examples of how to specify the Oracle customizer harness debug
option to run AuditorInstaller
in its default mode:
% sqlj -P-debug Foo_SJProfile0.ser Bar_SJProfile0.ser % sqlj -P-debug *.ser % sqlj -P-debug myappjar.jar
The debug
option results in the customizer harness instantiating and invoking the following class:
% sqlj.runtime.profile.util.AuditorInstaller
This class performs the work of inserting the debugging statements.
The -P-debug
option is equivalent to the following:
-P-customizer=sqlj.runtime.profile.util.AuditorInstaller
This overrides the customizer specified in the SQLJ -default-customizer
option.
Be aware of the following:
To run an application with auditors installed, the SQLJ file translator.jar
(or .zip
) must be in your classpath. (Normally, running a already translated SQLJ application requires only a runtime library.)
As with any Oracle customizer, help output and an option list will be provided if you specify -P-debug
together with -P-help
on the SQLJ command line.
It is important to realize that because the debug
option invokes a customizer, and only one customizer can run in any single running of SQLJ, you cannot perform any other customization when you use this option.
You also cannot use more than one of -P-print
, -P-debug
, and -P-verify
simultaneously, because each of these invokes a specialized customizer.
During run time, debugging statements placed by AuditorInstaller
result in a trace of methods called and values returned. This happens for all profile layers that had debugging statements installed. There is no means of selective debug output at run time.
AuditorInstaller
output relates to profiles only; there is currently no mapping to lines in your original .sqlj
source file.
Following is a sample portion of AuditorInstaller
run-time output. This is what the output might look like for a SQLJ SELECT INTO
statement:
oracle.sqlj.runtime.OraProfile@1 . getProfileData ( ) oracle.sqlj.runtime.OraProfile@1 . getProfileData returned sqlj.runtime.profile.ref.ProfileDataImpl@2 oracle.sqlj.runtime.OraProfile@1 . getStatement ( 0 ) oracle.sqlj.runtime.OraProfile@1 . getStatement returned oracle.sqlj.runtime.OraRTStatement@3 oracle.sqlj.runtime.OraRTStatement@3 . setMaxRows ( 1000 ) oracle.sqlj.runtime.OraRTStatement@3 . setMaxRows returned oracle.sqlj.runtime.OraRTStatement@3 . setMaxFieldSize ( 3000 ) oracle.sqlj.runtime.OraRTStatement@3 . setMaxFieldSize returned oracle.sqlj.runtime.OraRTStatement@3 . setQueryTimeout ( 1000 ) oracle.sqlj.runtime.OraRTStatement@3 . setQueryTimeout returned oracle.sqlj.runtime.OraRTStatement@3 . setBigDecimal ( 1 , 5 ) oracle.sqlj.runtime.OraRTStatement@3 . setBigDecimal returned oracle.sqlj.runtime.OraRTStatement@3 . setBoolean ( 2 , false ) oracle.sqlj.runtime.OraRTStatement@3 . setBoolean returned oracle.sqlj.runtime.OraRTStatement@3 . executeRTQuery ( ) oracle.sqlj.runtime.OraRTStatement@3 . executeRTQuery returned oracle.sqlj.runtime.OraRTResultSet@6 oracle.sqlj.runtime.OraRTStatement@3 . getWarnings ( ) oracle.sqlj.runtime.OraRTStatement@3 . getWarnings returned null oracle.sqlj.runtime.OraRTStatement@3 . executeComplete ( ) oracle.sqlj.runtime.OraRTStatement@3 . executeComplete returned oracle.sqlj.runtime.OraRTResultSet@6 . next ( ) oracle.sqlj.runtime.OraRTResultSet@6 . next returned true oracle.sqlj.runtime.OraRTResultSet@6 . getBigDecimal ( 1 ) oracle.sqlj.runtime.OraRTResultSet@6 . getBigDecimal returned 5 oracle.sqlj.runtime.OraRTResultSet@6 . getDate ( 7 ) oracle.sqlj.runtime.OraRTResultSet@6 . getDate returned 1998-03-28
There are two lines for each method call. The first shows the call and input parameters; the second shows the return value.
Note:
The classes you see in theoracle.sqlj.runtime
package are SQLJ run time classes with equivalent functionality to similarly named JDBC classes. For example, OraRTResultSet
is the SQLJ run time implementation of the JDBC ResultSet
interface, containing equivalent attributes and methods.As with any customizer, AuditorInstaller
has its own options that can be set using the -P-C
prefix on the SQLJ command line (or profile.C
in a SQLJ properties file).
AuditorInstaller
supports the following options:
depth
: Specify how deeply you want to go into the layers of run time functionality in your profiles.
log
: Specify the target file for run time output of the debugging statements of the installed auditor.
prefix
: Specify a prefix for each line of run time output that will result from this installation of debugging statements.
showReturns
: Enable the installed auditor to include return arguments in its run time call tracing.
showThreads
: Enable the installed auditor to include thread names in its run time call tracing (relevant only for multithreaded applications).
uninstall
: Remove the debugging statements placed into the profiles during the most recent previous invocation of AuditorInstaller
on those profiles.
As discussed in "Overview of Auditors and Code Layers", AuditorInstaller
can install a set of debugging statements, known as an auditor, at only a single layer of code during any one execution. The AuditorInstaller
depth
option enables you to specify which layer. Use multiple executions of AuditorInstaller
to install auditors at different levels.
Layers are numbered in integers. The shallowest depth is layer 0; a maximum depth of 2 or 3 is typical. The only depth settings typically used are 0 for the shallowest layer or -1 for the deepest layer. In fact, it is difficult to install an auditor at any other particular layer, because the layer numbers used for the various kinds of SQLJ executable statements are not publicized.
The depth
option is sometimes used in conjunction with the prefix
option. By running AuditorInstaller
more than once, with different prefixes for different layers, you can see at runtime what information is coming from which layers.
If you do not set the depth
option, or the specification exceeds the number of layers in a given profile, then an auditor will be installed at the deepest layer.
Use the log
option to specify an output file for runtime output that will result from the auditor that you are currently installing. Otherwise, standard output will be used, so that debug output will go to wherever SQLJ messages go.
When auditors write messages to an output file, they append; they do not overwrite. Therefore, you can specify the same log file for multiple auditors without conflict. In fact, it is typical in this way to have debug information from all layers of your application go to the same log file.
Use the prefix
option to specify a prefix for each line of runtime output resulting from the debugging statements installed during this invocation of AuditorInstaller
.
This option is often used in conjunction with the depth
option. By running AuditorInstaller
multiple times with different prefixes for different layers, you can easily see at runtime what information is coming from which layers.
Use the showReturns
option to enable or disable the display of return arguments as part of the runtime call tracing. This is enabled by default.
The following few lines show sample output with showReturns
enabled (default):
oracle.sqlj.runtime.OraRTStatement@3 . executeComplete ( ) oracle.sqlj.runtime.OraRTStatement@3 . executeComplete returned oracle.sqlj.runtime.OraRTResultSet@6 . next ( ) oracle.sqlj.runtime.OraRTResultSet@6 . next returned true oracle.sqlj.runtime.OraRTResultSet@6 . getBigDecimal ( 1 ) oracle.sqlj.runtime.OraRTResultSet@6 . getBigDecimal returned 5 oracle.sqlj.runtime.OraRTResultSet@6 . getDate ( 7 ) oracle.sqlj.runtime.OraRTResultSet@6 . getDate returned 1998-03-28
With showReturns
disabled, the output would appear as follows:
oracle.sqlj.runtime.OraRTStatement@3 . executeComplete ( ) oracle.sqlj.runtime.OraRTResultSet@6 . next ( ) oracle.sqlj.runtime.OraRTResultSet@6 . getBigDecimal ( 1 ) oracle.sqlj.runtime.OraRTResultSet@6 . getDate ( 7 )
Instead of both a call line and a return line for each method call, there is only a call line.
Use the showThreads
option to enable or disable the display of thread names as part of the runtime call tracing (relevant only for multithreaded applications). This is disabled by default.
When this option is enabled, thread names prefix the method names in the trace output.
Use the uninstall
option to remove debugging statements placed during previous invocations of AuditorInstaller
. Each time you use the uninstall
option, it will remove the auditor most recently installed.
To remove all auditors from a profile, run AuditorInstaller
repeatedly until you get a message indicating that the profile was unchanged.
Following are some full SQLJ command-line examples showing the specification of AuditorInstaller
options.
Insert a set of debugging statements, or auditor, into the deepest layer (which is the default layer), with runtime output to standard output:
% sqlj -P-debug MyApp_SJProfile*.ser
Insert an auditor into the deepest layer, with runtime output to log.txt
:
% sqlj -P-debug -P-Clog=foo/bar/log.txt MyApp_SJProfile*.ser
Insert an auditor into the deepest layer, with runtime output to standard output, showing thread names but not return arguments:
% sqlj -P-debug -P-CshowThreads=true -P-CshowReturns=false MyApp_SJProfile*.ser
Insert an auditor into layer 0 (the shallowest layer). Send runtime output to log.txt
; prefix each line of runtime output with "Layer 0:
" (the following command is a single wraparound line):
% sqlj -P-debug -P-Clog=foo/bar/log.txt -P-Cdepth=0 -P-Cprefix="Layer 0: " MyApp_SJProfile*.ser
Uninstall an auditor (this uninstalls the auditor most recently installed; do it repeatedly to uninstall all auditors):
% sqlj -P-debug -P-Cuninstall MyApp_SJProfile*.ser