Please refer to the Instantiations Smalltalk web page for technical information, including tips, and product updates made after this product release. The web page includes information about what's included in the release, and how to contact us as well as hints and tips for using and deploying VA Smalltalk. You can download the latest product updates from the Instantiations Smalltalk web site.
Installing VA Smalltalk For installation information, see cd_m\doc\instgd.htm or cd_c\doc\instgd.htm in the CD drive or the temporary directory where you extracted the manager or client installation files. The Installation Guide is also available at Instantiation's Smalltalk website. To install VA Smalltalk, follow the instructions for your specific platform. Before installing Version 7.0, follow these steps: 1. If EMSRV is running from the directory to be updated, stop EMSRV. 2. Stop any running VA Smalltalk images.
If you have a version of VisualAge Smalltalk installed, please refer to the Migration Guide for important information before using VA Smalltalk Version 7.0. The Migration Guide can be installed with the VA Smalltalk Client by selecting the "VA Smalltalk Documentation" feature.
The following sections list some important information about some of the components and features. For the latest product information, please refer to Instantiations Smalltalk web page.
***** - versioning/releasing owned parts in Organizer causes a walkback
***** - Add tracing to the SCI DLL on Windows platforms
VA Smalltalk lets programmers create and deploy e-business applications that are cross-platform and object-oriented: New support for Red Hat Linux, SUSE and Fedora
Be sure that your numlock key is turned off if you are using the composition editor. The numlock will prevent parts from being dropped on the Composition Editor.
Each time the Slider part is repainted the "X Error: BadWindow (invalid Window parameter)" message is printed in the xterm window which VisualAge Smalltalk was launched. The Slider part still functions normally.
On UNIX, make sure your LC_MONETARY locale setting contains a non-empty mon_decimal_point entry. On some machines, mon_decimal_point may be empty for the "C" locale. For example, if you wish to change to the en_US locale, set your LANG environment variable to en_US before starting VA Smalltalk with the ksh command: export LANG=en_US You can check the value of mon_decimal_point with the command: locale -k LC_MONETARY The output should look like: int_curr_symbol="USD " currency_symbol="$" mon_decimal_point="." mon_grouping="3" mon_thousands_sep="," positive_sign="" negative_sign="-" int_frac_digits=2 frac_digits=2 p_cs_precedes=1 p_sep_by_space=0 n_cs_precedes=1 n_sep_by_space=0 p_sign_posn=1 n_sign_posn=1 debit_sign="DB" credit_sign="CR" left_parenthesis="(" right_parenthesis=")"
The Object>>#-> method has been added as a convenience for constructing instances of Associations (for example, evaluating the expression 'upperLeft' -> (0@0) will answer anAssociation with 'upperLeft' as the key and 0@0 as the value). Some add-on products have extended Object with this same method. One such product is the RefactoringBrowser. Attempting to load such a product will cause a conflict and result in a load failure with a message similar to the following: Error: 330 Cannot complete the load because Object>>#-> is defined by RBParserVAApp and CLDT NOTE: The method collisions will be resolved if you reload after executing: EmImageBuilder cancelIfMethodsCollide: false Following the suggested course of action will allow the add-on product to load correctly.
Support for ANSI Smalltalk (see ANSI/NCITS 319-1998 Smalltalk Programming Language, available in PDF format from http://www.techstreet.com/cgi-bin/detail?product_id=56122) is included in this release of VisualAge Smalltalk Enterprise. This support greatly enhances the portability of Smalltalk applications between different Smalltalk implementations that provide ANSI Smalltalk support. All methods supporting the ANSI Smalltalk protocol are categorized as ANSI-API. Methods associated with ANSI Smalltalk function that is not complete in this release are categorized as ANSI-Unimplemented. The following restrictions with respect to ANSI Smalltalk are in force for this release: The following Exception methods are not functional isNested outer pass resignalAs: retry retryUsing: The following DateAndTime methods are not functional: timeZoneAbbreviation timeZoneName The MessageNotUnderstood exception class is provided, but is not signaled by the standard #doesNotUnderstand: method. The ZeroDivide exception class is provided, but is not signaled when a divide by zero occurs.
As a result of the changes made by the fix to defect 19427 in the EsString>>#bindWithArguments: processing (which also affects all EsString>>#bindWith:..... methods), documentation for the handling of missing or extra arguments is modified, and documentaiton for the pre-existing handling of the %0 sequence in the template string is added. These changes will be reflected in the IBM Smalltalk Programmers Reference when it is republished. Update IBM Smalltalk Programmers Reference, Chapter 13: NLS, section "Tools for developing International software", sub-section "Using message templates": 1) In the paragraph beginning "A template string is expanded...", delete the last sentence. 2) Add a new paragraph following this paragraph: There are also 2 special escape sequences. The double percent escape sequence ('%%') is replaced by a single percent chatacter in the composite message. The percent zero escape sequence ('%0') is replace by a platform line delimiter in the composite message. 3) in the paragraph beginning with "A template string permits ...", replace the last sentence with: Argument that are not referenced in the template string are ignored. Template string references to missing arguments are replaced by the escape sequence itself. 4) Repace the last entry in the "Resultant message" column of Table 49 with the following: 'Missing arguments are %2.'
Under certain Linux configurations, some operations that use progress dialogs can cause empty or "phantom" dialog boxes to remain after the operation has completed. These phantom dialog boxes can show up as a small rectange that is completely blank and cannot be moved. To work around this problem, make the following modification to EtWindow>>#execLongOperation:message:allowCancel:showProgress: ... aBlock argumentCount = 1 ifTrue: [runBlock := [aBlock value: inProgressDialog]] ifFalse: [runBlock := [(Delay forMilliseconds: 100) wait. aBlock value]]. ...
On MVS, an .ini file is optional. On all other platforms, an .ini file is required. The .ini file may have the same name and be in the same directory as your executable (on Unix, the executable is es or esnx). The .ini file can also have the same name and be in the same directory as your .icx or .ic file. In addition, you can specify your .ini file as a command line parameter. For example, you can launch your program by typing the following: abt -imyapp.icx -ini:c:\any.ini
On OS/2, if you have an older version of VisualAge Smalltalk running, when you attempt to start VisualAge Smalltalk 6.0, you will fail with the following message: "The program in session encountered a problem. Registry: the system could not demand load the application's segment ABT->ESVM40. EsReportWarning is in error. Help Sys127." To workaround this problem, start VA Smalltalk 6.0 before starting the older version. Another solution is to replace the bin directory of your older version with a copy of the bin directory that is installed with VisualAge Smalltalk 6.0.
When your development image is running on UNIX, you will not be able to read stack dump files that were generated on an architecture with different endian-ness. Doing so will cause a walkback. To work around this problem, swipe and execute the following doit from your development environment: PlatformLibrary mapLogicalName: 'EsLoadAndSave' toPhysicalName: 'libeslsi40'
Problem: Resumable execption instance variable not set (should answer an empty collection) Exception withAllSubclasses select: [:each | each exceptionalEvent resumable isNil] defaultHandler instance variable not set (should answer an empty collection?) Exception withAllSubclasses select: [:each | each exceptionalEvent defaultHandler isNil] 3rd (and after) element, of the contents of ExceptionalEventCollection, is the Exception subclass and not an instance of ExceptionalEvent (should answer false) (MessageNotUnderstood, Error, Warning) ancestorOf: Notification exceptionalEvent Sample below should answer nil (same as Notification signal) but instead executes the exception block although Notification does not inherit from Error (should answer nil). [Notification signal] on: Error do: [:error | error exitWith: 'Not Okay'] Examples: All the following (non comment) lines should answer true when evaluated. "Class based exception problems" ([Notification signal. 'Hello' ] on: Warning, ZeroDivide, Notification do: [:error | error resume: 'Goodbye']) = 'Hello'. ([Notification signal. 'Hello' ] on: Warning, ZeroDivide, Notification do: [:error | error return: 'Goodbye']) = 'Goodbye'. ([ZeroDivide signal. 'Hello' ] on: Warning, ZeroDivide, Notification do: [:error | error resume: 'Goodbye']) = 'Hello'. ([ZeroDivide signal. 'Hello' ] on: Warning, ZeroDivide, Notification do: [:error | error return: 'Goodbye']) = 'Goodbye'. ([Notification signal. 'Hello'] on: Error do: [:error | error resume: 'Goodbye']) = 'Hello'. ([Notification signal. 'Hello'] on: Error do: [:error | error return: 'Goodbye']) = 'Hello'. ([Notification signal. 'Hello'] on: Warning do: [:error | error resume: 'Goodbye']) = 'Hello'. ([Notification signal. 'Hello'] on: Warning do: [:error | error return: 'Goodbye']) = 'Hello'. ([Object new not foo. 'Hello'] on: Error do: [:error | ^true]). ([Object new not foo. 'Hello'] on: Error do: [:error | 'Goodbye']) = 'Goodbye'. ([Object new not foo. 'Hello'] on: Error do: [:error | error return: 'Goodbye']) = 'Goodbye'. ([Object new not foo. 'Hello'] on: Error do: [:error | error resume: 'Goodbye']) = 'Hello'. ([Notification signal. 'Hello'] on: Notification do: [:error | error pass]) = 'Hello'. ([Warning signal. 'Hello'] on: Notification do: [:error | error resume: 'Goodbye']) = 'Hello'. ([Warning signal. 'Hello'] on: Notification do: [:error | error return: 'Goodbye']) = 'Goodbye'. Solution: The ANSI exception handling code has been reworked as a result of this defect. 2 aspects of this code update may cause customers who work outside the scope of the identified API method some concern: 1) It was discovered that an erroneous Exception class>>#signal: API method was provided. This protocol is not defined by the ANSI Specification. Therefore, although the method was left in place, it was commented appropriately and recategorized to 'ANSI-Obsolete' meaning that its use is depricated. 2) The atttempt to homogonize the old and new forms of exception handling simply won't work. Undoing this failed attempt results in the following changes: -- The creation and extension of ExceptionEventCollection for class-based exceptions is removed and replaced by the creation and extension of a new ExceptionSet class. Since there was no explicit API for creating instances of these collections, this should be a generally transparent change. In addition, all instance-side ANSI-API has been moved from ExceptionalEventCollection to ExceptionSet since ExceptionalEventCollection no longer supports holding ANSI-style exceptions. -- The Block>>#on:do: method is changed to handle ONLY ANSI-style class-based exceptions. Handling the old style instance based exceptions was an undocumented extension of function in this method that is now eliminated.
The fix for APAR PQ62316 changed the algorithm used to hash a ScaledDecimal object. This change affects existing collections that rely on the hash value. The affected collections are, at least, KeyedCollection (and its subclasses) and Set (and its subclasses). If your application uses ScaledDecimal objects as keys for any of these hashed collections, you MUST recompute the hash values after loading the fix and before modifying the collection in any way. To recompute the hash values, send the rehash message to the collection (for example, myDictionaryKeyedWithScaledDecimals rehash).
Due to a subtle change to the compiler, some methods, when recompiled, may behave differently in 6.0.1 and 6.0.2 than they did in earlier versions. The original problem involved an error in the way the compiler determined whether 2 array literals were identical when the array literals contained one or more numeric values. See badArrayIdentity below as an example. This problem was corrected in 6.0.1, but some unforseen side-effects were introduced as shown in the table below. These side-effects have been eliminated in 6.0.2. While testing the changes for 6.0.2, we discover and fixed another (pre-existing) problem in this same code as demonstrated by poolDictionaryModification below. badArrayIdentity | a | a := #(3.0). ^a == #(3) "Should answer false" stringLiteralIdentity | a | a := 'hello'. ^a == 'hello' "Should answer true" mixedArray |a| a := #('string' #symbol 3). ^a == #('string' #symbol 3) "Should answer true" poolDictionaryModification "Upon method compilation, both constants = 'y'" "Upon entry to this method, TestPoolB::Value1 = 'y'" TestPoolA::Value1 := 'x'. ^TestPoolB::Value1 "Should answer 'y'" 6.0.0 6.0.1 6.0.2 badArrayIdentity true false false stringLiteralIdentity true false true mixedArray true false true poolDictionaryModification 'x' 'x' 'y'
Support for socket options get/set via the setsockopt and getsockopt function calls, previously available only for TCP stream and datagram sockets, has been added for Local Sockets on Linux/Unix platforms in V 6.0.1. There are two basic types of options: boolean options that enable/disable specific flags and others that set/get configurable features. Here are the names of the options that are typically supported by the underlying TCP/IP stack: SODEBUG, SOACCEPTCONN, SOTYPE, SOLINGER, SOREUSEADDR, SOKEEPALIVE, SOSNDBUF, SODONTROUTE, SOBROADCAST, SOERROR, SODONTLINGER, SORCVBUF. The operations use a boolean for the following options: SOACCEPTCONN, SODEBUG, SOREUSEADDR, SODONTROUTE, SOKEEPALIVE, SOBROADCAST, SODONTLINGER. The operations use an integer for the following options. SOERROR, SOSNDBUF, SOTYPE, SORCVBUF. The operations use a two entry array for the SOLINGER option. The first entry is a boolean and the second is an integer. In addition, a pair of connected local sockets can now be created via the SciSocketManager>>socketpair: call. For example, to create a pair of connected local stream sockets, call SciSocketManager default socketpair: SciSocketConstants::SOCKSTREAM. A collection containing a pair of connected local stream sockets will be returned ready for use.
By default, the send and receive buffers used for local sockets are set at 32 K (32768 bytes). In addition, there are platform-dependent limitations the programmer should be aware of. On Linux, the size of messages sent over a local socket should not exceed 63 KB, (64,512 bytes). On Solaris, the message size should not exceed 64 KB (65536 bytes). On HP-UX, the message size should not exceed 32 KB (32768 bytes). Finally, at the time of this writing, the limits on AIX are 2 KB (2048 bytes) for local datagram sockets and 4 KB (4096 bytes) for local stream sockets.
Support for MQCONNX and MQBEGIN function calls, along with their associated options and structures has been updated. MQCONNX is similar to MQCONN except that it allows options to be specified that controls the way the call works. The AbtMQBOStruct allows the application to specify options relating to the creation of a unit of work. The structure is an input/output parameter on the MQBEGIN call. The AbtMQBOStruct is not available for MQ clients. Refer to MQSeries documentation for further information about these functions and structures.
When loading the MQSeries feature on MVS, users will notice the following warning... Warning: 91 Defined global MQConnectX in unmanaged namespace while loading AbtMQCall class>>connectXTo:with:. This error is due to the existence of the MQConnectX function in the AbtMQSeriesBaseApp while not being defined as it should in hte AbtMQSeriesMVSSubApp. Although this problem will not break applications created with versions of VisualAge Smalltalk up to and including V 6.0, this function will not be available to new applications unless defined. To avoid this error message a definition for the function can be placed in AbtMQSeriesMVSSubApp class>> _PRAGMA_AbtMQSeriesFunctionsMVS by adding the following code to the pragma declaration... (name: MQConnectX isConstant: true valueExpression: '(PlatformFunction callingConvention: #c function: 'MQCONNX' library: nil parameterTypes: #(#pointer #pointer #pointer #pointer #pointer) returnType: #void)')
Support has been added for MQSeries on Linux and HP-UX platforms.
VisualAge Smalltalk Enterprise V6.0 does not support APPC and CPIC on Solaris, HP-UX or Linux.
There are several additons to SSL support in version 6.0.2 for peer X509 certificate verification on the SciSslsocket and SciSslSocketConfiguration levels. The information in this README item is meant to complement the Secure Sockets Layer chapter of the Smalltalk Programmer's Reference. An example SciSslSocketConfiguration using this new verfication API appears at the end of this README. SciSslSocket protocols - Instance Methods sslGetPeerCertificate Returns a pointer to the X509 certificate the peer presented or an SslError. If the peer did not present a certificate, nil is returned. Due to the protocol definition, a TLS/SSL server will always send a certificate, if present. A client will only send a certificate when explicitly requested to do so by the server. If an anonymous cipher is used, no certificates are sent. sslSetVerify: verifyMode Set the verification mode for the SSL connection. Valid modes of operation are... SSL_VERIFY_NONE - Server - the server will not send a client certificate request to the client, so the client will not send a certificate. Client - if anonymous ciphers are not in use (by default disabled), the server will send a certificate which will be checked. The result of the certificate verification process can be checked after the TLS/SSL handshake using the sslVerifyCertificate method. The handshake will be continued regardless of the verification result. SSL_VERIFY_PEER - Server - the server sends a client certificate request to the client. The certificate returned (if any) is checked. If the verification process fails, the TLS/SSL handshake is immediately terminated with an alert message containing the reason for the verification failure. The behaviour can be controlled by or-ing the additional SSL_VERIFY_FAIL_IF_NO_PEER_CERT and SSL_VERIFY_CLIENT_ONCE flags. Client - the server certificate is verified. If the verification process fails, the TLS/SSL handshake is immediately terminated with an alert message containing the reason for the verification failure. If no server certificate is sent, because an anonymous cipher is used, SSL_VERIFY_PEER is ignored. SSL_VERIFY_FAIL_IF_NO_PEER_CERT - Server - if the client did not return a certificate, the TLS/SSL handshake is immediately terminated with a handshake failure alert. This flag must be used together with SSL_VERIFY_PEER. Client - ignored SSL_VERIFY_CLIENT_ONCE - Server - only request a client certificate on the initial TLS/SSL handshake. Do not ask for a client certificate again in case of a renegotiation. This flag must be used together with SSL_VERIFY_PEER. Client - ignored Note: Exactly one of the mode flags SSL_VERIFY_NONE and SSL_VERIFY_PEER must be set at any time. Verification failure will lead to a termination of the TLS/SSL handshake with an alert message, if SSL_VERIFY_PEER is set. sslSetVerifyDepth: Set the maximum depth to which the certificate chain will be verified. This depth is only used if SSL_set_verify has previously been set to SSL_VERIFY_PEER. sslVerifyCertificate Answers the result of peer certificate verification. If verification mode is not set to SSL_VERIFY_NONE (see sslSetVerify:), sslVerifyCertificate checks the validity of the peer's X509 certificate to the specified verify depth, or 1 level if unspecified. Although there are many ways the verification of a certificate can fail (expired, revoked, invalid, etc.), the only error code that is returned is the last error that occurred during processing. SciSslContext protocols - Instance Methods certificateChainFile: aCertificateChainFilename The certificate must be in PEM format and must be sorted starting with the subject's certificate (actual client or server certificate), followed by intermediate CA certificates if applicable, and ending at the highest level (root) CA. certificateFile: anX509CertificateFilename Sets the context's certificate to be anX509CertificateFilename. The file must be in PEM format. privateKey: anSslPrivateKey Sets the privateKey used by this context to encrypt and decrypt data flowing over the SSL/TLS connection. sslMethod Answers the SSL method being used by this context. The SSL method describes the protocol version being used by the peer using this context. sslMethod: anSslMethod Sets the SSL method being used for this connection to be the method specified by aPlatformFunction. SSLV2 - SSL version 2 SSLV3 - SSL version 3 SSLV23 - SSL version 2 or 3 TLSV1 - TLS version 1 (sometimes called 3.1). SciSslSocketConfiguration protocols - Instance methods caFile: aCertificateAuthorityFile Sets the CA file for the configuration. The Certificate Authority (CA) file is a trusted certifcate used for verification purposes. caFile Answers the CA file for the configuration. caPath: aCAPath Sets the path to the directory containing CA certificate files. caPath Answers the path to the directory containing CA certificate files certificateChainFilename: aPemCertificateChainFilename Sets the certificate chain file to be used by the configuration for verification. certificateChainFilename Answers the certificate chain file beign used by the configuration for verification. verify: verifyMode Set the verification mode for the SSL configuration. Valid modes of operation are... SSL_VERIFY_NONE - Server - the server will not send a client certificate request to the client, so the client will not send a certificate. Client - if anonymous ciphers are not in use (by default disabled), the server will send a certificate which will be checked. The result of the certificate verification process can be checked after the TLS/SSL handshake using the sslVerifyCertificate method. The handshake will be continued regardless of the verification result. SSL_VERIFY_PEER - Server - the server sends a client certificate request to the client. The certificate returned (if any) is checked. If the verification process fails, the TLS/SSL handshake is immediately terminated with an alert message containing the reason for the verification failure. The behaviour can be controlled by or-ing the additional SSL_VERIFY_FAIL_IF_NO_PEER_CERT and SSL_VERIFY_CLIENT_ONCE flags. Client - the server certificate is verified. If the verification process fails, the TLS/SSL handshake is immediately terminated with an alert message containing the reason for the verification failure. If no server certificate is sent, because an anonymous cipher is used, SSL_VERIFY_PEER is ignored. SSL_VERIFY_FAIL_IF_NO_PEER_CERT - Server - if the client did not return a certificate, the TLS/SSL handshake is immediately terminated with a handshake failure alert. This flag must be used together with SSL_VERIFY_PEER. Client - ignored SSL_VERIFY_CLIENT_ONCE - Server - only request a client certificate on the initial TLS/SSL handshake. Do not ask for a client certificate again in case of a renegotiation. This flag must be used together with SSL_VERIFY_PEER. Client - ignored Note: Exactly one of the mode flags SSL_VERIFY_NONE and SSL_VERIFY_PEER must be set at any time. Set the verification mode for the SSL connection. Valid modes of operation are... verify Answer the verification mode for the SSL configuration. verifyDepth: anInteger Set the maximum depth to which the certificate chain will be verified to anInteger. This depth is only used if #verify: has previously been set to include SSL_VERIFY_PEER. verifyDepth Answer the maximum depth to which the certificate chain will be verified The following is an example SciSslSocketConfiguration using the new API to enable peer certificate verification... SciSslSocketConfiguration new certificateFilename: '<path>/certificateFilename.pem'; privateKeyFilename: '<path>/privateKeyFilename.pem'; caFile: '<path>/caFilename.pem'; sslVersion: SciSslConstants::SSLv3; verify: SciSslConstants::SSL_VERIFY_PEER | SciSslConstants::SSL_VERIFY_FAIL_IF_NO_PEER_CERT; verifyDepth: 1; yourself.
The Secure Socket Layer (SSL) support for VisualAge Smalltalk uses OpenSSL binaries. OpenSSL is an open source implementation of SSL/TLS based on the SSLeay library developed by Eric A. Young and Tim J. Hudson. The use of OpenSSL is provided under a dual license, the OpenSSL license and the SSLeay license. A copy of this license can be found at the end of Chapter 13. SSL in the Programmer's Reference. The SSL feature is supported on Windows 2000/NT/XP/98/ME, Linux, AIX, HP-UX, and Solaris. The SSL feature is not supported on MVS and OS/2 platforms.
When using stored procedures with the new Oracle 8 database connection, the Get schema function on the Stored Procedure Specification Settings view only works for procedures that are not contained in packages. Users must manually define host variables for procedures that are contained in packages.
ODBC drivers have not been shipped with VisualAge Smalltalk since Version 4.5. The drivers in previous versions were provided by MERANT (formerly INTERSOLV Inc.). If you need to obtain ODBC drivers, the DataDirect product is still available directly from MERANT. For more information call 800-876-3101 or visit http://www.merant.com/datadirect You can also check your DBRM for ODBC drivers. Most, if not all, major DBRMs now ship with ODBC drivers.
Before you can run the database features on AIX, you must extract a shared object from the appropriate archive file. This is true for IBM DB2, ODBC, and Oracle databases. IBM DB2 For IBM DB2, extract the file from $DB2INSTANCE/sqllib/lib/libdb2.a by performing the following steps: 1.Extract the shared object ar -x libdb2.a 2.Rename the extracted file libdb2.so mv shr.o libdb2.so ODBC For ODBC, extract from your libodbc.a file by performing the following steps: 1.Extract the shared object ar -x libodbc.a 2.Rename the extracted file libodbc.so mv libodbc.o libodbc.so Oracle For ORACLE, extract from your libclntsh.a file by perform the following steps: 1.Extract the shared object ar -x libclntsh.a 2.Rename the extracted file libclntsh.so mv clntsh.o libclntsh.so Note: For each of these databases, the resulting .so file must be in the library path (LIBPATH) in order to be located by VisualAge.
On HP and Solaris, the abt script file attempts to set up the shared library path to include DB2 if DB2 is detected. However, the Korn shell on HP and Solaris does not always evaluate the tilde character (~) so that VisualAge can set up the shared library path to include DB2. This causes the libraries for DB2 to not be added to the path correctly. To workaround this problem, you must add the DB2 libraries. You may want to add the one of the following examples to your profile. HP: export SHLIB_PATH=/home/db2inst1/sqllib/lib:$SHLIB_PATH Solaris: export LD_LIBRARY_PATH=/home/db2inst1/sqllib/lib:$LD_LIBRARY_PATH
On Unix, if you are using database features and experience a core dump when exiting VisualAge, comment out the PlatformLibrary>>shutDown method. An alternative solution for your packaged application is to execute the following code when exiting: System primitiveExit
Prior to this fix, the only provided method for changing the Oracle initialization mode was to create a subclass of AbtOracle8DatabaseManager in order to override the #userDefinedInitializeMode method. While this method is still available, it is now possible to change the default initialization mode by setting a class variable. The provided method for doing this is AbtOracle8DatabaseManager class>>#defaultInitializeMode: . The shipped default value is OCI_DEFAULT, which is unchanged from previous VisualAge Smalltalk version. See the comment in AbtOracle8DatabaseManager class>>#defaultInitializeMode: for other possible values.
Some database applications will need to add a prerequisite for the AbtRecordStructureApp application. Applications that use Database Parts will not need to add this prerequisite because the parts will include the AbtRecordStructureApp application. If an application manipulates instances of any of the subclasses of AbtRow, they will probably need to add this prerequisite. If you package your application and get the error The attribute Pub <Attr name> does not exist at runtime, you need to include the AbtRecordStructureApp application.
To create the stored procedure used in the Oracle sample, logon to SQL*PLUS and use the file found in your vast\samples\oracle. 1. Logon to SQL*PLUS sqlplus scott/tiger 2. Execute the file SQL> @vast\samples\oracle\sample.sql
When installing on a DBCS machine please remember the following: 1.If you are using OS/2 Warp 4.0J, you must apply Warp J4 Fixpack FX00004 before using VisualAge Smalltalk. 2.If you are using a DBCS version of OS/2 Warp 4.0, other than OS/2 Warp 4.0J, IBM VisualAge Smalltalk Enterprise requires the equivalent to Warp 4.0J Fixpack FX00004. 3.If you are using IBM VisualAge Smalltalk Enterprise on a DBCS system, you must use the ABTRULES.DBC file instead of the default US-English ABTRULES.NLS provided by the system. The ABTRULES.DBC file contains additional codepage conversion tables needed for the DBCS environment. You can find this file in your VisualAge NLS directory. Back up the ABTRULES.NLS file to ABTRULES.BAK, then rename ABTRULES.DBC to ABTRULES.NLS.
To ensure all information is displayed on your computer, we encourage you to use the highest resolution offered by your display terminal.
The NLS versions of Lotus Notes must be installed on the native Operating System (OS) platforms, in order for Notes to work. If a US-English version of Lotus Notes is installed on the native OS, then the user will not be able to input either SBCS or DBCS characters correctly. This is a restriction with Lotus Notes.
DBCS cookies are not supported using the Servlet Interface. This is a limitation of the HTTP Server.
On the UNIX platform, if you use Netscape 4.7, VisualAge Smalltalk may not be able to bring up Netscape when you try to access the help system. To workaround this problem on HP-UX and Sun Solaris, bring up Netscape manually before accessing the VisualAge Smalltalk Help. On AIX, bring up Netscape manually and type in the following URL : http: //localhost:57002/abtwebx/6.0/va/vast.htm
The VisualAge Smalltalk Domino Connection feature is now supported on Windows XP when running with Lotus Notes V6.0.
When detaching a file attachment using the Domino connection, the detached file is corrupted. The problem only shows up if the attachment is BASE64 encoded. Detaching the same attachment using the Notes Client works fine. Detaching via Domino connection also works fine if the file has been attached manually using the Notes Client.
EMSRV 7.1a and EMADMIN 7.0a are made available with this modification level. The changes from EMSRV 7.1 and EMADMIN 7.0 are as follows: Platform Support -- the following additional operating system levels are supported by EMSRV 7.1a: Windows XP Professional Solaris 8.0 NetWare 6.0 Bug fixes: Fixed bug in EMADMIN where a copy of a library would fail with an error if the size of the library in bytes was exactly divisible by 32768. Fixed bug in EMSRV where in extremely rare conditions, connections with locks might be incorrectly terminated by the "EMSRV inactivity monitor". Such terminations would be accompanied by a warning message in the log but since the default behavior of EMSRV is only to report errors, these messages were not often seen.
In an OS/2 environment, if EMSRV is installed under a directory name that contains spaces (e.g. x:\VAST60 MG\bin), attempting to invoke EMSRV via an OS/2 START command may fail with a SYS1041 message. For example, when issued from the varoot\bin directory, the following command: START EMSRV -u <userid> <password> may get the message SYS1041: The name EMSRV is not recognized as an internal or external command, operable program or batch file. To bypass this problem, issue the command sequence without the START: EMSRV -u <userid> <password>
Be sure to use the following good development practices with EMSRV: - Backup the manager file regularly. - Run library statistics utilities on a regular basis to ensure the integrity of the manager file. - Protect the manager file.
Running any release of EMSRV for Windows NT/2000/XP on a machine with more than one processor may lead to code repositories becoming corrupt. The increasing number of reports of corrupt code repositories resulting from running EMSRV in these environment has caused the following changes to be made to EMSRV: EMSRV 7.0a (shipped only with VisualAge Java) and EMSRV 7.1 (shipped with VisualAge Smalltalk 5.5 and 6.0) -- if EMSRV detects more than one installed processor, it will report the following to the console (or to the Application Log if EMSRV is running as a service) and then exit: WARNING: Running EMSRV for Windows NT/2000 on multiprocessor hardware is not supported due to the likelihood of a repository becoming corrupted. EMSRV 7.1a (shipped with VisualAge Smalltalk 6.0.1) -- if EMSRV detects more than one installed processor, and the -mp command line option is not specified, it will report the following to the console (or to the Application Log if EMSRV is running as a service) and then exit: WARNING: Running EMSRV for Windows NT/2000 on multiprocessor hardware is not supported due to suspected operating system bugs that may result in repository corruptions. Install and run EMSRV on a machine with a single processor or start EMSRV with the -mp option to bypass this check. The check for SMP hardware can be bypassed by starting EMSRV with the -mp command line option. By doing this, you will be running EMSRV on an unsupported platform and must assume full responsibility if code repositories become corrupted. When starting with the -mp command line option, EMSRV will still report a warning to the console (or to the Application Log if EMSRV is running as a service): WARNING: Running EMSRV for Windows NT/2000 on multiprocessor hardware is not supported due to suspected operating system bugs that may result in repository corruptions. You have chosen to start EMSRV with the -mp option to bypass a check that normally restricts EMSRV from running on multiprocessor hardware. This may cause repositories to become corrupted. Note that EMSRV continues to support SMP hardware in all other operating system environments.
If you are attempting to access the VisualAge Smalltalk newsgroup news://news.software.ibm.com/ibm.software.vasmalltalk using a Netscape browser, you must choose one of the following items in the Netscape browser's Edit->Preferences->Advanced->Proxies menu: 1. Direct connection to the Internet 2. Manual proxy If you have selected an autoproxy from this menu, your attempt to access the VisualAge Smalltalk newsgroup will fail.
On an AIX machine, before installing Smalltalk for the first time, use Smitty or Smit to create a Journalled File System.
On an AIX machine, before installing base Smalltalk for the first time, use Smitty or Smit to increase the disk size to 200 Megabytes.
We have seen rare cases where the client installation program on Linux will hang and become unresponsive after only partially completing the install. We have only noticed this on machines whose hard drive is one giant partition located at /. If this happens, a safe workaround is to terminate the install program and completely delete everything in the /opt/IBMvast/6.0 directory. Then restart the install.
On some Unix platforms, problems have been reported on the Select Features screen. After selecting a new feature to install, sometimes the Next button is not enabled due to an error in synchronization of the features. To correct this problem, select the Back button and then on the License Agreement screen, select the Accept button. The Next button on the Select Features screen should then be enabled.
Use copy and paste to share OLE objects between the Windows Explorer and an OLE Client part. Dropping an OLE object that was dragged from the Windows Explorer onto an OLE Client part does not work.
The fix to APAR PQ62227 introduced a change in packaging behavior for ICs. Previously, the #initializeClassVariable:to:inObjectNamed: and #initializeClassVariable:toObject:inClassNamed: packaging rules were being ignored. After the fix is applied, these packaging rules have their desired effect of initializing the identified class variable(s) in the packaged runtime IC. You should examine the senders of these messages in your code to ensure that they specified the initialization values that you want for the class variables at runtime.
When using the Packager UI (Modifiy Instructions :: Applications and ICs), you must manually add AbtNlsCfsSupportApp to the image that you are packaging as follows: 1.Select AbtNlsCfsSupportApp in the left pane. 2.Press the >> button. (It will be highlighted. There are two of these buttons. You want to press the one on the left that is below the left and/or center panes).
The RMI Wizard adds the following instance methods to all mapped classes: sstRmiClassName Answers the Java class name the receiver is mapped to. sstIsRmiSerializable Answers true if the receiver is serializable (passed by value). sstIsRmiRemotable Answers true if the receiver is remotable (passed by reference). The above methods, along with the class mapping definitions (added to the application class), are used by SST to enable instances of the class for use with RMI. There may be some cases where you want to enable the class itself (versus instances of the class) for use with RMI. For example, you might want to have a Java client send messages to a Smalltalk class. If this is the case, you'll need to add the above methods as class methods.
An SST HTTP server now writes an "access" log, named "httpd.log". Entries in the log are formatted according to the default format used by the Apache HTTP server, in order to enable tooling for Apache logs to be used for SST logs as well. This access log is initialized in SstHttpServer>>startUp. An alternate formatter or message target can be specified by sending #accessLog: prior to #startUp.
The SST HTTP client support for HTTP proxies has been enhanced to provide for tunneling HTTPS through an HTTP proxy. The central feature of this enhancement is a new connection class - SstHttpsTunnelConnection - which knows how to do the upgrade from TCP to SSL. To make use of this enhancement, a client must use a transport configuration that specifies a #socketClass of SciSocket and a #connectionClass of SstHttpTunneledConnection. The following method has been added as a convenience for defining a transport configuration to support this scenario. SstHttpClient class>>#initializeTransportScheme:forHttpsTunnelThrough:proxyAuth: Example: SstHttpClient initializeTransportScheme: 'local_https_tunnel' forHttpsTunnelThrough: 'proxy.acme.com:80' proxyAuth: nil The method above accepts 'nil' as a proxyAuth credential to indicate that proxy authentication is not required. The class SstHttpClient is introduced with this enhancement. A client using the transport configuration defined above would be created using the factory method SstHttpClient class>>#forTransportScheme:, as in the example below. |client| (client := SstHttpClient forTransportScheme: 'local_https_tunnel') startUp. [client get: 'https://www.foobar.com/secure/index.html'] ensure: [client shutDown]. It is also possible to configure SST so that the convenience method SstHttpUrl>>fetch uses this tunneling feature. This is done in the same way as illustrated above for an application-specific transport configuration, specifying instead the identifer for the default https client configuration. SstHttpClient initializeTransportScheme: 'httpsl' forHttpsTunnelThrough: 'proxy.acme.com:80' proxyAuth: nil Having done this, the following code is equivalent to the example above. 'https://www.foobar.com/secure/index.html' sstAsUrl fetch
Typically, security policies indicate that an HTTPS client should validate the X509 certificate presented by an HTTPS server, and should verify that the identity asserted by a validated certificate is in fact the identity of the intended server. Version 6.0.2 includes enhancements to the SST HTTPS client which enable applications to specify via the security configuration for an HTTPS client transport that a strict certificate validation policy should be applied. In addition, an SstHttpClient can be configured with a @requiredPeerName that will result in the client auto-verifying the identity asserted by the server credential. To enable strict certificate validation, the HTTPS client transport security configuration must specify: 1) @caFile - the file name of a trusted certificate authority (CA) certificate store, and 2) @verify - the validation policy, and 3) @verifyDepth - the maximum depth of the presented certificate chain. The @verify parameter must be set to "SSL_VERIFY_PEER|SSL_VERIFY_FAIL_IF_NO_PEER_CERT". Unless known to be too limiting, a reasonable default value for @verifyDepth is the Integer '2'. The @caFile is a certificate store in PEM format, as per OpenSSL. Example 1. The code below configures the default HTTPS client transport for strict server certificate validation. (SstTransport configurationRegistry at: 'httpsl') securityConfiguration: (SciSslSocketConfiguration new sslVersion: SciSslConstants::SSLv3; caFile: 'certs/sst_ca.pem'; verify: SciSslConstants::SSL_VERIFY_PEER | Sc iSslConstants::SSL_VERIFY_FAIL_IF_NO_PEER_CERT; verifyDepth: 2; yourself). With this configuration in place, the fetch below will fall into the exception handler if the presented server certificate cannot be validated against the trusted CA certificate store. ['https://www.acme.com/secure/index.html' sstAsUrl fetch] when: SstConstants::ExSstNonFatalError do: [:sig| sig exitWith: ('SSL handshake failed:*' match: sig arguments last errorObject)]. To enable server identity verification, in addition to certificate validation, SstHttpsConnection now has a "getter" for @peerCredential, allowing an application to test the Subject attribute of the server's X509 certificate. Auto-verification is enabled in SstHttpClient via a new @requiredPeerName attribute. If this attribute is set, the client will test the certificate Subject immediately after successful SSL connection negotiation (prior to sending the HTTP request) and generate an exception if the Subject does not match the specified @requiredPeerName. Example 2. |client response| (client := SstHttpClient forTransportScheme: 'httpsl') startUp. client requiredPeerName: '/who/do'. [['https://www.acme.com/secure/index.html' sstAsUrl fetch] ensure: [client shutDown]] when: SstConstants::ExSstNonFatalError do: [:sig| sig exitWith: ('Invalid peer:*' match: sig arguments last errorObject)].
When running the IIOP PingPong examples you must pass a type that conforms to the CORBA Any type interface. Typically, Strings are used as the argument representing @anAnyType when sending the method SstPingPongIIOP>>start: @anInteger with: @anAnyTYpe. Passing Smalltalk Integers and Floats as arguments will cause the example to fail because CORBA does not represent these as objects, and therefore, they do not conform to the Any interface.
On AIX, some SST applications that use the MQ transport layer will fail when using the unthreaded versions of the AIX MQ libraries. If you are getting the error MqCallInProgress, this may be the cause. By default, Smalltalk MQ calls will use the unthreaded libraries. To switch to the threaded libraries, before making your first MQ call, execute the call AbtMQSeriesBaseUnixSubApp threaded.
In previous releases of VisualAge Smalltalk Server Runtime, only one method of logging a simple walkback was provided. When an error occurred, the walkback information was written to TranscriptTTY. This caused the walkback information to be written to the console (Unix) or to a log file identified by the -l commandline option. Since TranscriptTTY does unbuffered character-at-a-time output, it can be very time consuming to write the walkback information. For VisualAge Smalltalk V 5.5.2 and later, an alternative output mechanism is provided. When it is enabled, this mechanism writes the walkback information to a file stream just as would be done for a Reduced Runtime image. This is a buffered operation which can be more than an order of magnitude faster than writing to TranscriptTTY. To enable writing the walkback information to a file stream, you must provide the startUp class with the filename of the file to be associated with the file stream. For example, to see the difference in behavior, create an XD image and load the HelloWorld application. Then package it specifying AbtHeadlessRuntimeStartUp as the image startup class. If you specify HelloWorld haltHelloWorld as the application entry point, the walkback will be written to TranscriptTTY (you need to specify the -l commandline switch at runtime to see this output on Windows and OS/2); if you specify System image startUpClass walkbackFileName: 'runtimeWB.log'. HelloWorld haltHelloWorld as the application entry point, the walkback will be written to the runtimeWB.log file.
CORBA IIOP facilities provided by SST in previous releases are obsolete as of this release. The implementation provided in previous releases continues to be shipped with this release, but all methods have been recategorized as 'OBSOLETE'. There will be no further developement or enhancement of these facilities, and they may be removed from the product in a future release. Customer applications which made use of these facilities in a previous release will continue to be supported as they are migrated to this current release. Customers are advised to make use of Web Services technologies, such as XML and WSDL, for future interoperability strategies.
The RMI in Server Smalltalk will run under JDK 1.3, using the same techniques that were required for JDK 1.2.
SstUnixSocketDemultiplexer>>sstWarn: sends #abtPadWith:upToLength:onRight:, which is implemented in AbtCLDTAdditions - an app not in the prereq chain of the controller for SstUnixSocketDemultiplexer>>sstWarn:. This method will likely be removed during packaging, causing a runtime DNU. Workaround: include AbtCDLTAdditions when packaging an SST app.
When trying to use the interactive debugger, if you are getting the error EHOSTNOTFOUND or EADDRNOTAVAIL on the runtime side, the problem may be that your runtime machine cannot resolve the dotted TCP/IP address of your development machine. You can work around this problem by adding an entry to the hosts file on the runtime machine for your development machine.
When a Table part is dropped onto an Html Page, the #selectionType attribute for a table column contains the following selections in the properties table: "<Error: No Form - Multiple Select>" " Error: No Form - Single Select>" These messages are a bit confusing. The #selectionType attribute is only valid for tables that are dropped onto an Html Form part.
It is not currently possible to use the interactive debugger facility of an XD image to debug a Web Connection application. In order to debug a Web Connection application (or an XML application that uses the Web Server Interface), perform the following steps: 1.In the AbtWebServerInterfaceBaseApp>>AbtWsiConnection>>#handleTransaction: method, change the ExAll exception reference to ExError. 2.In the AbtWebServerInterfaceBaseApp>>Block>>#abtWsiAtEndOrWhenExceptionDo: method, replace the code with the following: abtWsiAtEndOrWhenExceptionDo: completionBlock " Code hacked to enable debugging in XD runtime image via the interactive debugger " self value. completionBlock value After making the above changes, package the application and execute it as usual. Note: Be sure to load the original code prior to packaging the application for production.
The Web Server Interface Monitor is no longer included, by default, in the prerequisites for Web applications. In order to include the monitor in a packaged application, users should modify the prerequisites for web applications to include theAbtRunHtmlPageApp application. Alternatively, users can package their web applications using the Tools->Browse Packaged Images option. You can add the AbtRunHtmlPageApp application to the packaged image from the Package Control Panel without modifying application prerequisites. This approach is necessary for Web applications that must be loaded into an XD image because the AbtRunHtmlPageApp application will not load into an XD image. For example, the sample application AbtWebSamplesApp is now headless by default because it does not include the prerequisite AbtRunHtmlPageApp. When the packaged image for this sample is started, no windows will open. The application AbtWebSamplesApp can be loaded into a passive XD image and packaged if desired. The application AbtWebSamplesWithMonitorUIApp contains prerequisites to include the Web Server Interface Monitor as well as all the sample parts from AbtWebSamplesApp. This application cannot be loaded into an XD passive image. Note: Applications constructed before v4.5 already include the prerequisite AbtRunHtmlPageApp, so no special action is necessary.
On OS/2, attempting to start a WSI Server with transport type sst-http causes Smalltalk to hang if TCP/IP loopback is not enabled. Eventually, a Smalltalk debugger appears with the following error message: Could not create socket pair: ETIMEDOUT (10060): Connection timed out. To correct this problem, enable loopback on OS/2 by performing the following steps: 1.Open TCP/IP Configuration windows. 2.Select loopback interface from the Interface to Configure list box. 3.Select the Enable interface check box. 4.Close the TCP/IP Configuration window. To determine if the loopback interface is working properly, type the following from an OS/2 command prompt: ping 127.0.0.1 If loopback is properly configured, a series of messages will be written to the OS/2 session indicating that the target address for the ping was successfully contacted. Press ctrl-c to terminate the ping operation. After successfully configuring loopback, you should be able to use the sst-http interface from OS/2.
To package your Web Connection application so that it utilizes the Web Connection image components, you must implement a packager method to force inclusion of your web parts in the packaged image. For example, implement the following method as a class method of the application containing your web connection parts. packagerIncludeClassNames ^self defined collect: [:i | i name ]
Several new runtime files are required by the VAST Web services feature for 6.0.2. They are: sstaxis.xsd - An XML schema that contains definitions for common Java types including 'Map', 'Hashtable', and 'Vector' sstaxis.map - An XML mapping specification that enables seamless mapping of common Java types into Smalltalk objects. abtvast.map -- Contains unique mappings to enable unambiguous serialization of objects that can be mapped to a variety of schema types.
The Web services examples have been augmented to demonstrate new Web services functionality (WS-I Basic Profile. SOAP 1.2, security enhancements). The directory structure for the samples has been modified. See the file ./samples/sstws/Readme.txt for additional information about using the Web services samples.
The targetNamespace for the implementation WSDL of the insurance example was changed from "http://www.SstWSInsurancePolicyInterface.com/SstWSInsurancePolicyInterface" to "http://www.SstWSInsurancePolicyInterface.com/SstWSInsurancePolicyInterface-impl" (added '-impl' suffix). The documenation still references the old namespace in some of the code samples; however, the code samples in the .\samples\sstws directory are correct.
When a Web services client receives a SOAP fault response from the server, standard Web services handler chains are traversed prior to signalling an exception to report the SOAP fault. Therefore, fault chains are traversed after standard client processing has already completed. Custom client chains may need to be aware of the possibility that the message result could be a fault. Ideally, the Web services support should circumvent standard processing and branch immediately into the fault handlers after discovering a SOAP fault in the output message. Client handlers can use the message #isSoapFault to verify that message results contain expected content. example: fault := anSstWSMessageContext results detect:[:each | each isSoapFault] ifNone: [ ].
------------------------------- How can I override the default XML serialization logic? XML serialization is performed by objects called 'serializers'. Developers can supply custom serializers that replace the default serialization logic. | container config | container := SstWSContainer containerNamed: 'MyContainer'. config := container serializationManager serializationConfigurationNamed: SstSoapConstants::SstSoapDefaultEnvelopeNamespace. config serializer: MyCustomSerializer current. Alternatively, the serialization configuration itself can be replaced. | container | container := SstWSContainer containerNamed: 'MyContainer'. container serializationManager addSerializationConfiguration: AbtXmlSerializationConfiguration newSoapConfiguration named: SstSoapConstants::SstSoapDefaultEnvelopeNamespace. ------------------------------- How can I disable/enable SOAP multi-ref serialization? By default, SOAP encoded messages are constructed using the serialized named SstSoapOutputSerializer. This serializer causes all complex SOAP messages to be encoded as SOAP multi-refs. Notice that the serialization configuration is referenced using the SOAP encoding namespace. | container config | container := SstWSContainer containerNamed: 'MyContainer'. config := container serializationManager serializationConfigurationNamed: SstSoapConstants::SstSoapDefaultEncodingNamespace. config serializer: AbtXmlSchemaOutputSerializer current. ------------------------------- How can I supply custom deserializers to provide custom logic for mapping XML into objects? Mapping parsed XML into Smalltalk model objects is performed by objects called 'deserializers'. Developers can supply custom deserializers to enhance default mapping behavior. Custom deserializers can be created and set in deserialization configurations. | container config | container := SstWSContainer containerNamed: 'MyContainer'. config := container serializationManager deserializationConfigurationNamed: SstSoapConstants::SstSoapDefaultEnvelopeNamespace. config deserializer: AbtXmlSchemaInputDeserializer current. Alternatively, the default SOAP deserialization configuration can be replaced as shown below. | container config | container := SstWSContainer containerNamed: 'MyContainer'. config := AbtXmlDeserializationConfiguration newSoapConfiguration. config deserializer: AbtXmlSchemaInputDeserializer current. container serializationManager addDeserializationConfiguration: config named: SstSoapConstants::SstSoapDefaultEnvelopeNamespace. ------------------------------- How can I print incoming and outgoing SOAP messages on my Web services server? The message handler chain can be modified to add code that dumps the contents of the SOAP message and response. | factory pluggableHandler | factory := ( SstWSContainer containerNamed: 'VastSampleServerContainer' ) handlerFactory. pluggableHandler :=SstWSPluggableHandler new invokeBlock: [ :anSstWSMessageContext | | message | [ message := anSstWSMessageContext propertyNamed: ##wsTransportMessageRequest. AbtXmlUtility logError: 'Request->'. AbtXmlUtility logError: message header debugPrintString. AbtXmlUtility logError: message contents asString. AbtXmlUtility logError: ''. AbtXmlUtility logError: 'Response->'. " NOTE: The output message is not stored in the message context. " AbtXmlUtility logError: ( anSstWSMessageContext container serializationManager serialize: anSstWSMessageContext currentMessage context: anSstWSMessageContext )] fork ]. factory register: pluggableHandler named: 'wsHttpServerResponseHandler' inNamespace: factory globalNamespace To disable the tracing code, execute the snippet below: | factory | factory := ( SstWSContainer containerNamed: 'VastSampleServerContainer' ) handlerFactory. factory register: SstWSNoOperationHandler default named: 'wsHttpServerResponseHandler' inNamespace: factory globalNamespace ------------------------------- How can I print incoming and outgoing SOAP messages on my Web services client? The message handler chain can be modified to add code that dumps the contents of the SOAP message and response. | factory pluggableHandler | factory := ( SstWSContainer containerNamed: 'VastSampleClientContainer' ) handlerFactory. pluggableHandler :=SstWSPluggableHandler new invokeBlock: [ :anSstWSMessageContext | | message | [ message := anSstWSMessageContext propertyNamed: ##wsTransportMessageRequest. AbtXmlUtility logError: 'Request->'. AbtXmlUtility logError: message asString. AbtXmlUtility logError: ''. AbtXmlUtility logError: 'Response->'. " NOTE: The output message is not stored in the message context. " message := anSstWSMessageContext propertyNamed: ##wsTransportMessageResponse. AbtXmlUtility logError: message header debugPrintString. AbtXmlUtility logError: message contents asString ] fork ]. factory register: pluggableHandler named: 'wsHttpClientResponseHandler' inNamespace: factory globalNamespace To disable the tracing code, execute the snippet below: | factory | factory := ( SstWSContainer containerNamed: 'VastSampleClientContainer' ) handlerFactory. factory register: SstWSNoOperationHandler default named: 'wsHttpClientResponseHandler' inNamespace: factory globalNamespace ------------------------------- How can I clean up the sockets and active HTTP servers in my image? The script below shuts down all active HTTP servers, and terminates all active sockets. [true] ensure: ["Stop a running system and clean everything out" SstHttpServletEngine allInstances do:[:e | e shutDown. e clear. e become: nil]. SstHttpServer allInstances do:[:e | e shutDown. e clear. e become: nil]. Processor allProcesses do: [:s | (s == UIProcess currentUI) ifFalse: [s basicTerminate]]. Processor reschedule. SstApplicationContext clearAll. SciSocketManager default closeAllSockets.] ------------------------------- How does VAST default serializer determine the XML schema type for an object that is defined as 'anyType'? The algorithm for processing 'anyType' and for processing 'subtypes' is very similar. A new mapping spec is introduced (contained in 'abtvast.map') to represent the Smalltalk namespace. The new mapping spec is intended to describe how Smalltalk objects should map to XML schema types. For typical mapping specs, it is possible to have multiple types map to the same object; therefore, determination of a schema type from an object instance is ambiguous. The Smalltalk namespace mappings remove ambiguity because each object maps to a single schema type. For many cases, it is not necessary to refer to the Smalltalk mapping spec in order to determine a proper type mapping. The Smalltalk namespace exists strictly to remove ambiguity when the same class can map to multiple types. When trying to determine the 'actual' type for an object, VAST does the following: - Check the schema of the base schema type to see if there is a more specialized type that matches the class name of the object being serialized. - Try to find a mapping for the class name of the serialized object and use the mapping to determine the actual schema type. The mapping is searched for in: 1) Smalltalk mapping namespace 2) Base type namespace 3) All other visible namespaces (namespaces visible at that point in serialization of the object) For types other than 'anyType', if no 'specialized' type is found, the 'baseType' is assumed to be the serialization type. For an 'anyType', if no 'specialized' type is found, a type definition is constructed based on the class shape. The type is created in the "VisualAge Smalltalk" namespace. The "Smalltalk" namespace is settable in the serialization configuration. The default value is 'urn:Vast' and is globally settable in the pool variable AbtXmlConstants::AbtXmlSmalltalkNamespace. ------------------------------- How can I disable validation during WSDL parsing? AbtXmlObjectCache current addDeserializationConfiguration: (AbtXmlDeserializationConfiguration newWsdlConfiguration validate: false) named: 'http://schemas.xmlsoap.org/wsdl/'. ------------------------------- How can I enable code page conversion of SOAP messages during serialization? Add an 'XmlSerializationEncoding' entry to the [Xml] stanza of your application's .ini file as shown below. [Xml] DefaultResourceQualifier=D:\vast\b49\xml\ XmlSerializationEncoding=UTF-8 The value for this entry should be the target encoding for the serialized SOAP message. The VAST Web services support will always attempt to serialize SOAP messages using the specified encoding. For less common encodings, it may be necessary to add a code page mapping using API in AbtXmlStreamConverter to map the specified encoding value to a valid code page supported by the platform. AbtXmlStreamConverter mapEncoding: 'UTF-8' toCodePage: 65001 ------------------------------- How can I deploy commonly used schemas to a Web services container? The deployment descriptor contains a <schemaUrls> block for this purpose. Schemas that are imported by deployed WSDL documents are automatically deployed to the container and do not need to be included in the <schemaUrls> block of the deployment descriptor. <schemaUrls> <schemaUrl>http://schemas.xmlsoap.org/ws/2002/07/secext/</schemaUrl> </schemaUrls> Use the <schemaUrls> block to: - Specify XML schemas required to provide supplemental functionality (ie. SOAP 1.2, WS-Security ) - Override schemas that are already deployed to the global XML object cache (ie. WSDL, SOAP) ------------------------------- How can I enable a VAST container to process SOAP 1.2 messages? VAST Web services containers can be configured to enable processing of SOAP 1.2 messages. This is accomplished by deploying SOAP 1.2 schemas and mapping specifications to the server container. See the sample directory /samples/sstws/soap12/Readme.txt for additional information about configuring a Web services container for SOAP 1.2. The same Web services container can be used to process SOAP 1.1 and SOAP 1.2 messages. ------------------------------- How can I enable a VAST client service to send SOAP 1.2 messages? VAST Web services client containers can be configured to enable processing of SOAP 1.2 messages. This is accomplished by deploying SOAP 1.2 schemas and mapping specifications to the client container. In addition, SstWSService instances contain accessors to allow the SOAP 1.2 namespace to be specified for outgoing SOAP messages. ie) | service | service := (SstWSContainer containerNamed: 'MyContainer') serviceManager serviceNamed: 'SstWSInsurancePolicyInterface' inNamespace: 'http://www.SstWSInsurancePolicyInterface.com/SstWSInsurancePolicyInterface-impl'. service sstSoapEnvelopeNamespace: SstSoap12Constants::SstSoap12EnvelopeNamespace. See the sample directory /samples/sstws/soap12/Readme.txt for additional information about configuring a Web services container for SOAP 1.2. ------------------------------- How can I configure client services to pass WS-Security information? The schema (http://schemas.xmlsoap.org/ws/2002/07/secext/) must be managed by the serializationManager of the Web services container. This is accomplished by modifying the client deployment descriptor to include the required schema. ie) <schemaUrls> <schemaUrl>http://schemas.xmlsoap.org/ws/2002/07/secext/</schemaUrl> </schemaUrls> Authorization credentials can be specified in the SstWSService that is being invoked causing those credentials to be passed automatically as part of the message. The SstWSService contains accessors for getting and setting HTTP authorization credentials (#sstAuthCredential:) as well as WS-Security username tokens (#sstUsernameToken:). See the sample directory /samples/sstws/ws_security/Readme.txt for additional information and examples. ------------------------------- How can I specify a custom servlet for processing incoming service requests? The default servlet for processing incoming Web services messages is SstWSServlet. Developers may provide their own custom servlet if more sophisticated rocessing is required. To enable an alternative servlet using the default http transport strategy, execute code like that shown below: SstWSHttpNetworkTransportStrategy defaultServletClass: MyWebServicesServlet ------------------------------- How can I make use of custom SST transports for invocation of remote service operations? The container configuration supports transport mappings that are used to map url schemes to the registered SST transport schemes used for invocation. The #urlScheme setting can specify a full URL name to enable special processing for a specific resource location, or the value can be a scheme prefix (ie. 'http' or https') for general cases. The #transportScheme is the identifier for an SstTransport configuration that describes the details of the underlying communications protocol. <!-- transportMappings is an optional element that appears after the </managers> end tag in the container deployment descriptor. --> <transportMappings> <transportMapping urlScheme="http" transportScheme="httpl" /> <transportMapping urlScheme="http://vasthost:63003" transportScheme="vasthost" /> <transportMapping urlScheme="https" transportScheme="testhttps" /> </transportMappings> The code below accomplishes the same task: | myContainer | myContainer := SstWSContainer containerNamed: 'VastSampleClientContainer'. myContainer configuration mapUrlScheme: 'http' toTransportScheme: 'httpl' ; mapUrlScheme: 'http://vasthost:63003' toTransportScheme: 'vasthost' ; mapUrlScheme: 'https' toTransportScheme: 'testhttps' SST transport schemes must be registered in class SstTransport prior to usage. See the class methods #serverTransportConfiguration and registerPluginConfigurations in SstHttpCommunications for an example of how transport configurations are created and registered. ------------------------------- How can I map a Smalltalk Dictionary to XML? The resource files 'sstaxis.map' and 'sstaxis.xsd' contain the rules required for mapping Smalltalk Dictionary objects to/from XML. Dictionaries are represented in XML as 'Map' elements in the namespace 'http://xml.apache.org/xml-soap'
The XML server sample named AbtXmlSampleCustomerRequestHandler contains code with hard coded directory references that do not resolve properly. If you would like to test this sample, you must first do the following: 1.Create a subdirectory named xml from your VisualAge base client directory. For example: mkdir xml 2.Copy the xml files from the directory /opt/IBMvast/6.0/xml to the newly created xml directory. For example: cd <vast client directory> /xml cp /opt/IBMvast/6.0/xml/*.* .
If you wish to package your XML application so that it utilizes the XML image components, you must implement a packager method to force inclusion of your XML request handler parts in the packaged image. For example, you can implement the following method as a class method of the application containing your XML request handler parts: packagerIncludeClassNames | handlers | handlers := AbtXmlServerSamplesApp defined select: [:aClass | aClass inheritsFrom: AbtXmlWsiHandler ]. ^handlers collect: [:aClass | aClass name ] For reduced runtime images that are packaged without the XML image components, all XML request handlers are automatically included in the packaged image.
If the input serialization functionality of the XML feature is used to map incoming XML requests into Smalltalk business objects, the packager will exclude classes that are not specifically referenced in code. The specific sequence of events that may cause packaging concerns is as follows: 1.Parse an XML file to construct a DOM (document object model). 2.Send the #mapUsing: method to the DOM and pass it a valid instance of the AbtXmlMappingSpec class. The #mapUsing: API uses the passed mapping specification to build an object from the contents of the DOM. However, the steps above do not cause any actual references to the class name of the constructed instance. Therefore, packaging rules must be used to instruct the packager that unreferenced classes should be included in the packaged image. For example, the #packagerIncludeClassNames packager method (a class method) can be implemented in the application class. For example, if MyModel1 and MyModel2' are classes in MyApplication, then the method below should be implemented as a class method in MyApplication. packagerIncludeClassNames " Include class names that might be constructed via XML mapping " ^#(MyModel1 MyModel2)
The following public methods are removed: AbtXmlSchema>>#abtXmlPrintOn:namespaces: AbtXmlSchemaComplexType>>#abtXmlPrintOn:namespaces: AbtXmlSchemaDeclaration>>abtXmlPrintOn:namespaces: These methods were not used by supported versions of the VAST XML support and should not affect user applications.
This change does NOT affect objects that utilize an XML schema for serialization. Nil attribute values are now suppressed during serialization of attributes with attribute mappings that specify 'Required="false". For example: <ClassElementMapping ElementTagName="customer" ClassName="AbtXmlSampleCustomer" > <AttributeMapping ClassAttribute="lastName" Required="false"><Attribute>lastName</Attribute> </AttributeMapping> </ClassElementMapping> If the lastName attribute of a serialization target object is nil, the serializer will not render the 'lastName' attribute.
To try out the XML server samples, perform the following steps: Testing XML request handlers over HTTP 1.From the VisualAge Organizer, select the AbtXmlServerSamplesUIApp application. 2.Select the AbtXmlSampleHttpClientTesterView part. 3.Enter an XML string or a piece of code that evaluates to an XML string AbtXmlSampleCustomerRequest xmlTestString1 AbtXmlSampleCustomerRequest xmlTestString2 4.From the Actions menu, select the Open WSI monitor option to bring up the Web Server Interface monitor. Follow the instructions in the Web Connection User's Guide to start a WSI server with transport type wsi-tcp 5.Specifiy the URL that will handle the request: http://myserver/cgi-bin/abtwsac.exe/AbtXmlSampleCustomerRequestHandler 6.Enable the Options->Inspect result option so that you can view the returned XML response string 7.Select the code string in the text box, and select the Actions->Evaluate and send menu option (or use pop-up menu for text box). After the command is processed, an inspector should open to display an XML string that contains the results. Testing XML request handlers over sockets (very similar to the steps for testing over HTTP) 1.From the VisualAge Organizer, select the AbtXmlServerSamplesUIApp application. 2.Select the AbtXmlSampleSocketClientTesterView part. 3.Enter an XML string or a piece of code that evaluates to an XML string AbtXmlSampleCustomerRequest xmlTestString1 AbtXmlSampleCustomerRequest xmlTestString2 4.From the Actions menu, select the Open WSI monitor option to bring up the Web Server Interface monitor. Start a WSI server with transport type xml-tcp and select the request handler class that you wish to use for handling incoming requests on the socket. For example: AbtXmlSampleCustomerSaxRequestWsiHandler 5.From the XML socket client tester view, specify the server and port number for the request that is to be made (the same server and port from step #4 above). 6.Enable the Options->Inspect result option so that you can view the returned XML response string. 7.Select the code string in the text box, and select the Actions->Evaluate and send menu option (or use pop-up menu for text box). After the command is processed, an inspector should open to display an XML string that contains the results.
The XML parser automatically performs code page conversion before attempting to parse an XML stream. Many code pages are handled seamlessly using the default code page conversion routine of the runtime operating system. However, there are some character encodings that cannot be converted. Unsupported code page conversions cause walkbacks at execution time. The following code pages are not supported: * EUC_JP conversion is not properly supported by native Windows code page routines. A Windows implementation of ICONV supports EUC_JP. To download ICONV, see http://www.ibm.com/software/ad/smalltalk/downloads/vacodepage.html Using the ICONV support, additional code pages, including EUC_JP, can be supported. * The code page ISO-2022-JP is not supported by native routines or by ICONV. The VisualAge XML support attempts to map XML character set encodings to valid code pages. The default mappings can be overridden using the API shown in the following example: AbtXmlStreamConverter mapEncoding: 'UTF-8' toCodePage: 65001. VisualAge uses the code page conversion support APIs that are built in to each of the supported platforms. Therefore, code page mappings may be different for different operating systems. If you encounter a debugger with the following message it is likely that you have encountered an unmapped or mismapped encoding. Abt.Nls.160.e: Conversion from code page <sourceCodePage> to code page <targetCodePage> is not supported.
THIS DOCUMENT IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND. INSTANTIATIONS DISCLAIMS ALL WARRANTIES, WHETHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF FITNESS FOR A PARTICULAR PURPOSE AND MERCHANTABILITY WITH RESPECT TO THE INFORMATION IN THIS DOCUMENT. BY FURNISHING THIS DOCUMENT, INSTANTIATIONS GRANTS NO LICENSES TO ANY PATENTS OR COPYRIGHTS. (C) Copyright Instantiations Corporation 2005. All rights reserved.
[10/6/2003 - 11:19:02 PM]