cl_abap_typedescr=>describe_by_data->length reports double length !

[glio_ad]

Hello everybody and I wish you all a very Happy New Year !

Here is my problem:

I have created a function module with the following interface:

*"----

---

""Local Interface:

*" IMPORTING

*" VALUE(I_NAME) TYPE RVARI_VNAM

*" VALUE(I_TYPE) TYPE RSSCR_KIND

*" TABLES

*" T_RANGE

*" EXCEPTIONS

*" VAR_NOT_FOUND

*" NO_DATA

*"----

---

This FM will be used to read any variant maintained in table TVARV and will return its values in a given range (through the tables parameter t_range). For example, the program ZTEST will call this FM with a range r_ktokd (defined as: ranges: r_ktokd for kna1-ktokd) and I_NAME will have the value, say, 'ZSD_ACC_GRP'.

In order to be able to handle any type of range object defined in a program, I am trying to dynamically realize the actual length of the type range (in our example, KTOKD has length 4), so that I know exactly how to handle the data from TVARVC and move them respectively to t_range-low, t_range-high etc fields.

Therefore, I use have written the following code to do this:

DATA: wa_tvarvc TYPE tvarvc,

descr_ref TYPE REF TO cl_abap_typedescr,

wl_field_len TYPE i,

wl_high TYPE i.

CLEAR: wa_tvarvc, t_range, t_range[].

descr_ref = cl_abap_typedescr=>describe_by_data( t_range ).

• Calculate the low-high fields' length

wl_field_len = ( descr_ref->length - 3 ) / 2 .

as t_range-sign length = 1

t_range-option length = 2

t_range-low length = 4

t_range-high length = 4

the descr_ref->length should return 11 as the total length of the t_range structure. However, it returns 22 !! So, I have changed the code as follows:

wl_field_len = ( descr_ref->length - 6 ) / 4 .

Although it works now, can someone please explain me why this happens and if there is a more safe way to do this?

Thanks in advance and HAPPY CODING in 2008 !

George

[uwe_schieferstein]

Hello George

You have to go for the "full monty" as shown below in the sample report ZUS_SDN_DYNAMIC_TVARVC.

&----

---

*& Report ZUS_SDN_DYNAMIC_TVARVC *

*& *

&----

---

*& *

*& *

&----

---

REPORT zus_sdn_dynamic_tvarvc .

TYPE-POOLS: abap.

TABLES: kna1.

DATA:

gr_ktokd TYPE RANGE OF kna1-ktokd.

DATA:

gs_component TYPE abap_compdescr,

go_typedescr TYPE REF TO cl_abap_typedescr,

go_strucdescr TYPE REF TO cl_abap_structdescr,

go_tabledescr TYPE REF TO cl_abap_tabledescr.

START-OF-SELECTION.

CALL METHOD cl_abap_typedescr=>describe_by_data

EXPORTING

p_data = gr_ktokd " range = table type

RECEIVING

p_descr_ref = go_typedescr. " table type

" Casting to table type instance

go_tabledescr ?= go_typedescr.

" Get table line type

CALL METHOD go_tabledescr->get_table_line_type

RECEIVING

p_descr_ref = go_typedescr.

" Casting to line type instance

go_strucdescr ?= go_typedescr.

" Components of range

LOOP AT go_strucdescr->components INTO gs_component.

AT FIRST.

WRITE: /1 'Length',

AT 15 'Decimals',

AT 25 'Type',

AT 35 'Name'.

WRITE: / syst-uline.

ENDAT.

WRITE: /1 gs_component-length LEFT-JUSTIFIED,

AT 15 gs_component-decimals LEFT-JUSTIFIED,

AT 25 gs_component-type_kind,

AT 35 gs_component-name.

ENDLOOP.

END-OF-SELECTION.

Regards

Uwe

[glio_ad]

Hello Uwe and thanks for the reply.

I tried your code, however, gs_component-length keeps reporting double length for every component of the range structure!! I cannot figure out why this happens.... Might this be a Unicode issue??? a BASIS issue?? is it a bug or a feature? 😛

Anyway, thanks for your prompt answer.

Cheers,

George

[uwe_schieferstein]

Hello George

I modified my report in order to match your requirements more closely.

*&---------------------------------------------------------------------*
*& Report  ZUS_SDN_DYNAMIC_TVARVC                                      *
*&                                                                     *
*&---------------------------------------------------------------------*
*&                                                                     *
*&                                                                     *
*&---------------------------------------------------------------------*

REPORT  zus_sdn_dynamic_tvarvc                      .

TYPE-POOLS: abap.


TABLES: kna1.


DATA:
  gr_ktokd    TYPE RANGE OF kna1-ktokd.  " itab w/o header line

**DATA:
**  gs_component    TYPE abap_compdescr,
**  go_typedescr    TYPE REF TO cl_abap_typedescr,
**  go_strucdescr   TYPE REF TO cl_abap_structdescr,
**  go_tabledescr   TYPE REF TO cl_abap_tabledescr.


SELECT-OPTIONS:
  s_ktokd         FOR kna1-ktokd.  " itab with header line

START-OF-SELECTION.



  CALL FUNCTION 'Z_US_TEST1'
       TABLES
            xt_range = gr_ktokd.
  SKIP.


  CALL FUNCTION 'Z_US_TEST1'
       TABLES
            xt_range = s_ktokd.
  SKIP.

  CALL FUNCTION 'Z_US_TEST1'
       TABLES
            xt_range = s_ktokd[].
  SKIP.

And here is the coding of the function module:

FUNCTION z_us_test1.
*"----------------------------------------------------------------------
*"*"Lokale Schnittstelle:
*"  TABLES
*"      XT_RANGE
*"----------------------------------------------------------------------

  TYPE-POOLS: abap.

  DATA:
    gs_component    TYPE abap_compdescr,
    go_typedescr    TYPE REF TO cl_abap_typedescr,
    go_strucdescr   TYPE REF TO cl_abap_structdescr,
    go_tabledescr   TYPE REF TO cl_abap_tabledescr.


  CALL METHOD cl_abap_typedescr=>describe_by_data
    EXPORTING
      p_data      = xt_range
    RECEIVING
      p_descr_ref = go_typedescr.  " structure type, not table type !!!!

" Not required anymore
**  " Casting to table type instance
**  go_tabledescr ?= go_typedescr.
**
**  " Get table line type
**  CALL METHOD go_tabledescr->get_table_line_type
**    RECEIVING
**      p_descr_ref = go_typedescr.

  " Casting to line type instance
  go_strucdescr ?= go_typedescr.

  " Components of range
  LOOP AT go_strucdescr->components INTO gs_component.
    AT FIRST.
      WRITE: /1  'Length',
               AT 15 'Decimals',
               AT 25 'Type',
               AT 35 'Name'.
      WRITE: / syst-uline.
    ENDAT.

    WRITE: /1   gs_component-length LEFT-JUSTIFIED,
             AT 15  gs_component-decimals LEFT-JUSTIFIED,
             AT 25  gs_component-type_kind,
             AT 35  gs_component-name.
  ENDLOOP.

ENDFUNCTION.

Again the results are always correct and the same (lenght of LOW / HIGH = 4). This report was developed on 4.6c.

Regards,

Uwe

[glio_ad]

Hello Uwe.

Thank you for you prompt support! The last code you sent me is syntactically OK, but it keeps reporting double length!!

I tend to believe that it has to do with the fact that our system here (ECC 6.0) is Unicode...

Anyway, If anyone has some in-depth knowledge on this please advise!

Thanks again Uwe.

[Former Member]

Hello,

I'm having the same problem. Unfortunately, I cannot delete every value by 2, because not all field-lengths are doubled.

Has anyone figured out why this happens?

Kind regards,

Jorg