ABAP: Pass by Value vs. Pass by Reference

In modern ABAP Object Oriented Programming, we define method parameters using the IMPORTING, EXPORTING keywords, each of which can use pass by value or pass by reference, as well as the RETURNING keyword, which always uses pass by value, and the CHANGING keyword, which always uses pass by reference.

Let’s understand the difference between them.

1. Pass by Value

When a parameter is passed by value, a copy of the data is made.

This means any changes inside the method do not affect the original variable outside the method. In methods we explicitly define pass by value using the VALUE keyword.

Example 1) of Pass by Value

Explanation:

  • The VALUE keyword ensures iv_number is passed by value, thus a copy of the data is created inside the method.
  • Any modifications to the copied data within change_value do not affect lv_number outside the method.
  • Note: The IMPORTING parameter was able to be „changed“ due to VALUE( ) without needing to define it as a CHANGING parameter.

That’s great, but what if instead of:

We have:

We have IMPORTING VALUE(), which is implicitly pass-by-value, so it creates a copy of the data, but now we don’t have TYPE i but TYPE REF TO i as our type.

Let’s take a look.

Example 2) of Pass by Value

Explanation

  • lv_number is an integer and lo_number is a reference that points to its memory location.
  • The method is pass-by-value and expects a reference, so the reference (memory address) is copied.
  • When the dereferenced value behind that address is changed inside the method, the change persists outside the method even on the original lv_number since it shares the same address.
  • Note: lv_number->* ue be can be modified, despite lv_number being desfined as an IMPORTING parameter

Next, let’s check out pass-by-reference in methods.

2. Pass by Reference

When a parameter is passed by reference, no copy of the data is created.

Instead, the method copies the memory address of the data as a reference and operates directly on the original variable, meaning any changes made inside the method will reflect outside the method, as the original data was changed via this reference.

Example 1) of Pass by Reference

Explanation:

  • lv_number is passed by reference to the method, thus the method acts directly on the data behind the reference.
  • The changes to cv_number reflect outside the method on lv_number.
  • Note: cv_number must be made a CHANGING parameter, because IMPORTING parameters cannot be changed by reference.

Example 2) of Pass by Reference

Explanation:

  • lv_number is an integer and lo_number is a reference that points to its memory location.
  • The method is pass-by-reference and expects a reference, so a reference to a reference is acted upon.
  • When the dereferenced value behind that reference is changed inside the method, the change persists outside the method even on the original lv_number as it was passed by reference

3. Field-Symbols

We’ve pretty much covered pass-by-value and pass-by-reference in the context of methods and their parameters, and we’ve also talked about variables and ref variables.

I know some readers are going to be wondering: “What about Field-Symbols?”

So let’s briefly discuss them as well and you’ll find more details in the performance comparisons below.

What are field-symbols?

Field-symbols are pass-by-reference. Their syntax differs a bit from references, but basically they’re pointers in ABAP, allowing dynamic access to memory locations without copying data or needing to be dereferenced.

While field-symbols offer a slight performance advantage due to not needing to be dereferenced, which is a negligible performance cost, they are becoming obsolete and are not recommended by SAP.

Hence, why I only want to mention them briefly.

4. Compare & Contrast

Pass-by-value vs. Pass-by-reference:

AspectPass-by-valuePass-by-reference
ConceptThe method receivs a copy of the variables’s valueThe method receives a pointer (reference) to the original variable
Memory UsageHigher (creates a copy)Lower (uses original memory)
PerformanceSlower for large dataFaster, especially for tables
ScopeChanges only affect the copy within the methodChanges affect the original variable directly
UsageCommon for small data types or when memory is no concernCommon for large structures, itabs and objects
Data IntegritySafer (no unintended modifications)Risk of unintended modifications
ModificationsNo impact on the original dataModifies the original data directl.y
FlexibilitySafer, as the original data remains unchangedMore flexible, but requires careful handling
SyntaxIMPORTING VALUE( )
TYPE
IMPORTING
TYPE REF TO

Field-Symbols vs. References:

AspectField-Symbols (< >)References (REF #( ))
ConceptActs like a pointer, referencing existing memory directlyCreates a reference object pointing to data
Memory UsageNo new memory allocationRequires memory for the reference itself
PerformanceVery fast (direct access)Slightly slower due to dereferencing
ScopeOnly valid within the block where assignedPersistent across method calls
UsageGood for internal table processing, loops, dynamic accessUseful for object-oriented design, method parameters
ModificationsChanges the original data directlyChanges data via the reference
FlexibilityMore flexible for direct memory operationsBetter for passing objects and structured data
DereferencingImplicit (no extra step needed)Explicit (lo_ref->*)

*Dereferencing is very fast, operating at memory access speed. The performance cost is negligible compared to large data copies or database accesses.

5. Overview of Behaviour

  1. In ABAP OOP, using VALUE() or not in your IMPORTING and EXPORTING parameters determines whether the method implicitly behaves as pass-by-value or pass-by-reference.
  2. A reference to an object is the memory address of an object via which the object can be accessed (dereferenced, ->*) and acted upon. Passing only memory addresses requires less computation than copying the entire datasets around.
  3. Pass-by-value methods create a copy of the object, even when the object itself is a reference. Thus the original object stays unchanged outside the method.
  4. Pass-by-reference methods access the memory address of the original object and may act upon its data by dereferencing. The original object doesn’t stay unchanged outside the method. This can sometimes lead to unexpected behavior, when appending to EXPORTING tables in separate method calls throughout a program, where one would expect the parameter to be cleared automatically with each call, but isn’t. Manual clearing of EXPORTING parameters for tables, is often wanted, when appending to pass-by-reference parameters.
  5. CHANGING is always pass-by-reference and RETURNING is always pass-by-value. IMPORTING and EXPORTING can be either.

Conclusion

Modern ABAP OOP syntax allows clear control over pass by reference vs. pass by value, influencing performance and data integrity. By understanding these principles, developers can write more efficient and maintainable code in ABAP.

Let’s Keep the Conversation Going!

Do you have questions or your own experiences on this topic? Feel free to reach out to Jonathan Rumpl on LinkedIn!

Stay up to date! Sign up for our Cadaxo newsletter and receive regular insights, best practices, and news from the SAP world.

 

UI5 und der Cache

UI5/Fiori und der Cache!

Viele kennen die Situation. Man hat im Entwicklungssystem eine super coole UI5/Fiori App für ein S/4HANA System gebastelt, aber nach dem Import in das Produktivsystem (oder Q-System) geht erstmal gar nix. Target was not found, Type Error, Unknown Setting, Cannot read properties, …

Sehr oft liegen die Gründe der Probleme am Cache! Nicht nur am Browser Cache, auch im SAP Backend gibt es mehrere zu berücksichtigende Caches. Nachfolgend eine Auflistung der Cache Themen, die mir im Zuge meiner Arbeit untergekommen sind.

Eigentlich unglaublich, wie viele Caches man ggf. löschen / invalidieren muss.

Wichtiger Hinweis: Das sind meine persönlichen Erfahrungen, bitte verwendet die Reports oder Transaktionen nur nach vorheriger Prüfung. Es kann durchaus sein, dass der eine oder andere Report obsolet wird oder in gewissen System nicht verwendet werden darf! Ich übernehme keine Garantie, dass die Reports oder Transaktionen in euren Systemen einwandfrei und korrekt funktionieren.

NameTypErläuterungBereich
/UI5/APP_INDEX_CALCULATEReportAb UI Add-On 2.0+ Sollte periodisch eingeplant seinUI / App
/UI2/DELETE_CACHEReportLöscht UI2 Services & Target Mapping CachesUI / App / Lpd
/IWFND/CACHE_CLEANUPTransaktionClean Frontend OData CacheGW
/IWBEP/CACHE_CLEANUPTransaktionClean Backend OData CacheGW
ICM Server CacheGlobal InvalidierenUI / App / GW
/UI2/INVALIDATE_CLIENT_CACHESReportClient Cache Invalidation Fiori Launchpad / Cache BusterUI / App
/UI2/INVALIDATE_GLOBAL_CACHESReportInvalidiert alle UI2 Caches (Alle Services mit /UI2/*, SHMM, … )UI / App
/UI5/UPD_ODATA_METADATA_CACHEReportOData Cache – Cache Buster TokensUI / App
/UI5/UPDATE_CACHEBUSTERReportNur für UI Add-On 1.0UI / APP
CL_SADL_LOAD_AREAShared MemoryShared Memory invalidierenUI / APP
/UI2/FLP_DEL_PERSONALIZATIONReportSAP Fiori Launchpad PersonalisierungenUI / APP
/UI2/PAGE_CACHE_SYNCHRONIZE
/UI2/SYNC_PBC
Report
Transaction
Synchronization of the User-Independent CacheUI / APP
/UI2/EAM_BUILD_CACHEReportApp Finder Performance CacheUI / APP
Browser CacheIE, Chrome, … Cache UI / APP

/UI5/APP_INDEX_CALCULATE

Das ist mein Lieblingsreport in dem Zusammenhang, da er mir in den meisten Fällen immer die Probleme löst. Es gibt einen Applikationsindex und Caching für Apps, Komponenten und Libraries. Dieser Report löscht bzw. invalidiert diese Caches / Indextabellen und ist im Normalfall automatisch eingeplant.

Ich habe jedoch die Erfahrung gemacht, dass nach einem Transport von Änderungen einer UI5 Anwendung, ein explizites Ausführen des Reports für die jeweilige Anwendung notwendig ist.

Weiter Informationen zu dem Report können hier entnommen werden:

/UI2/DELETE_CACHE

Hier geht es um das Löschen des UI2 Cache. Dieser kann für einen User oder generell gelöscht werden.

/IWFND/CACHE_CLEANUP und /IWBEP/CACHE_CLEANUP

Invalidiert / Löscht das oData Model (SEGW oder RAP) am Frontend- und Backend-Server.

/UI2/INVALIDATE_CLIENT_CACHES

Dieser Report sollte überhaupt nur in Ausnahmefällen notwendig sein – wenn der UI5 Cache Buster im Einsatz ist. Eine Erklärung zum Report gibt es unter help.sap.com.

/UI2/INVALIDATE_GLOBAL_CACHES

Invalidiert alle globalen UI2 Caches. Bitte die Dokumentation des Reports im System beachten.

/UI5/UPD_ODATA_METADATA_CACHE

Der Report löscht die OData Caches Tokens. Bitte auch hier die Dokumentation des Reports beachten.

/UI5/UPDATE_CACHEBUSTER

Der sorgt dafür, dass der Cache Buster mit dem aktuellsten UI5 Metadaten arbeitet. Der Report ist jedoch nur bei Systemen mit UI Add-On 1.0 einzusetzen. Bei Systemen mit einem höheren UI Add-On Release ist hingegen /UI5/APP_INDEX_CALCULATE zu verwenden.

/UI2/FLP_DEL_PERSONALIZATION

Fiori Launchpad Homepages können normalerweise durch den Anwender personalisiert werden. Es können neue Gruppen angelegt werden, Apps entfernt oder hinzugefügt werden. Manchmal ist es notwendig, solche Personalisierungen zurückzusetzen. Beispielsweise nach Änderungen im Fiori Katalog oder einer Fiori Gruppe. Ein Erklärung zu der Funktion ist hier beschrieben.

/UI2/PAGE_CACHE_SYNCHRONIZE

Der Report bereinigt Cache Probleme im Bereich der Launchpad Catalogs/Groups. Laut SAP sollte der Report regelmäßig eingeplant werden. Der Report ist auch über die Transaktion /UI2/SYNC_PBC erreichbar.

/UI2/EAM_BUILD_CACHE

Hier geht es um die Performance im Bereich App Finder. Weiter Informationen zu dem Report können hier entnommen werden:

Sonstiges

Änderungen an Service-Definitionen müssen manchmal manuell in Transportaufträgen aufgenommen und transportiert werden.

Fazit

Es vergeht fast kein Transport wo nicht irgendein Cache ein Problem bereitet. Aber mit dem /UI5/APP_INDEX_CALCULATE bekommt man das meist in den Griff.

Ich werden diesen Blogpost nach und nach um weitere Erkenntnisse in dem Zusammenhang ergänzen.

How to send emails from ABAP® with CL_BCS

Usage of the class CL_BCS in order to send eMails from ABAP™ applications via the Business Communication Service.

With our development tips we address special, related to praxis topics around the exciting and inexhaustible topic software development.

Valid for: SAP NetWeaver ABAP Web AS from 6.10

Index

  • General
  • e-mail dispatch from SAP
  • Class CL_BCS
  • Method CREATE_PERSISTENT – compile of a sending task
  • Method SET_DOCUMENT – attach text, data files and other objects
  • Method ADD_ATTACHMENT of the class CL_DOCUMENT_BCS
  • Method ADD_RECIPIENT – transfer of receiver
  • Method SEND – Sending of e-mail
  • Cases of application
  • Copyright & disclaimer

General

Often it is important to have necessary information at disposal in order to react quickly and in the right way. Especially, when the information procurement, like a report, takes a long time and cannot be constantly watched, it is useful to release an action upon completion. That can for example be an eMail with the status of a batch job or even the ABAP list.

eMail dispatch from SAP

SAP provided with the web AS 6.10 (6.20) the Business Communication Services (BCS) to send objects from ABAP applications. This object orientated interface is the successor of the Business Communication Interface (BCI), which is implemented as objects of the Business Object Repository (BOR).

In this Cadaxo development tip I will give an overview of the e-mail dispatch via BCS.

Class CL_BCS

Via BCS different objects with attachments and notes for several receivers can be sent also in different communication ways. Consequently the interface – the class CL_BCS – is constructed complex. In this development tip we therefore concentrate on sending an e-mail with attachment only.

A sending process breaks down into the following steps:

  • Generation of a sending task
  • Attaching an object
  • Choice of receivers
  • Sending of the mail

For all those tasks the class CL_BCA provides the following methods:

  • CREATE_PERSISTENT Compilation of a sending task
  • SET_DOCUMENT Attaching a text, data or other objects
  • ADD_RECIPIENT Add the receiver
  • SEND Sending of e-mail

Method CREATE_PERSISTENT – Compilation of a sending order

This method generates a sending task and returns an object of the class CL_BCS. All further steps are used upon that object.

Method SET_DOCUMENT – Attachment oft ext, data and other objects

With this method the content, meaning the actual text and data attachments of the e-mail is committed. The method expects an object with the interface IF_DOCUMENTS_BCS. Attachments are committed via this object too.

Via the statistic method CREATE_DOCUMENT of the class CL_DOCUMENT_BCS a respective object can be created.

The method expects the following parameters:

I_TYPE: Type of committed data, e.g. HTM für eine HTML Datei, TXT for text data, BIN for binary data like PDFs

I_TEXT/I_HEX: A chart with the actual text resp. the binary data according to the choosen type

I_SUBJECT: subject of the e-mail

Method ADD_ATTACHMENT of class CL_DOCUMENT_BCS

This method is similarly used to the CREATE_DOCUMENT:

I_ATTACHMENT_TYPE: Type of committed data, e.g. HTM for HTML data, TXT for text data, BIN for binary data like PDFs

I_ATTACHMENT_TEXT/

I_ATTACHMENT_HEX: An internal table with the actuall text resp. the binary data according to the choosen type

I_ATTACHMENT_SUBJECT: Name of the attachment file

Method ADD_RECIPIENT – Transfer of receiver

Depending on the way of communication and type of the sent object different receivers can be transferred to the mail. E.g. no business object relation can be per e-mail.

The method expect at least one object which implements the interface IF_RECIPIENT_BCS:

I_RECIPIENT: receiver address (Interface IF_RECIPIENT_BCS)

I_EXPRESS: Express message; if the receiver is a SAP user, a pop-up appears at the next processing of the PBO event with the notice of a new message

I_COPY: The receiver receives the message as a copy

I_BLIND_COPY: The receiver receives the message as a blind copy

I_NO_FORWARD: no forwarding; if the receiver is a SAP user, he cannot forward the message

The two most important receiver types are probably e-mail addresses and SAP users. The respective objects are generated by the following statistic methods:

CL_SAPUSER_BCS=>CREATE For SAP users

CL_CAM_ADDRESS_BCS=>CREATE_INTERNET_ADDRESS For e-mail addresses

Method SEND – Sending of an e-mail

The message is passed on to SAPconnect and sent via the respective interface. The actual sending process is usually scheduled by a batch job with the report RSCONN01, which periodically sends all upcoming messages. In order to avoid the waiting perioduntil the next job performance, the message can be tagged as “to be sent immediately”: Method SET_SEND_IMMEDIATELY( abap_true ) of the sending task.

Method SHORT_MESSAGE

With this method the SAPOffice dialogue opens itself for new messages, on which the user can maintain receivers, texts, attachments and their settings himself. Via parameters suggestion values for the respective fields and objects, e.g. receiver addresses, can be taken with at the time of selection, which later will be displayed in the message.