CLASS ... DEFINITION DEFERRED not working

[ruhnla]

The ABAP cleaner stops immediately when the command CLASS ... DEFINITION DEFERRED is found with the error message "Opening 'CLASS' not closed." (ABAP version 7.50)

Is this a bug or a feature or missing functionality?

[jmgrassau]

Hi ruhnla,

that's odd, actually ABAP cleaner does consider CLASS ... DEFINITION DEFERRED and does not expect an ENDCLASS here. Could you maybe try to create a minimal example for this (removing everything that doesn't trigger the error)? Do you use the plug-in or the stand-alone ABAP cleaner, and which version of it?

Kind regards,
Jörg-Michael

[ruhnla]

Hi Jörg-Michael,

I have tried both versions of ABAP cleaner, ADT and standalone, latest version (1.71). In the standalone version I can't even paste the code.

Here you have a simple example of coding where it happens.

`

•   Class definition for ALV toolbar
  

CLASS: lcl_alv_toolbar DEFINITION DEFERRED.

DATA : c_alv_toolbarmanager TYPE REF TO cl_alv_grid_toolbar_manager,
c_alv_toolbar TYPE REF TO lcl_alv_toolbar.

•   CLASS lcl_alv_toolbar DEFINITION
  

CLASS lcl_alv_toolbar DEFINITION.
PUBLIC SECTION.

•   Constructor
  

  METHODS: constructor IMPORTING io_alv_grid TYPE REF TO cl_gui_alv_grid.

ENDCLASS. "lcl_alv_toolbar DEFINITION

•   CLASS lcl_alv_toolbar IMPLEMENTATION
  

CLASS lcl_alv_toolbar IMPLEMENTATION.

METHOD constructor.

•   Create ALV toolbar manager instance
  

  CREATE OBJECT c_alv_toolbarmanager
  EXPORTING
  io_alv_grid = io_alv_grid.
  ENDMETHOD. "constructor

ENDCLASS. "lcl_alv_toolbar IMPLEMENTATION
`

Best,
Lars

[jmgrassau]

Hi Lars,

I see – it's the colon! I didn't expect CLASS to be followed by :, so if you remove that, things will work (and I shall enhance the code to consider this case, too).

Kind regards,
Jörg-Michael

[jmgrassau]

Hi Lars,

okay, that wasn't difficult, this will be considered in the next release.

Kind regards,
Jörg-Michael

[ConjuringCoffee]

: seems to be one of the greatest threats to the ABAP Cleaner 😄

[jmgrassau]

Hi ConjuringCoffee,

yes, along with comments and pragmas that's indeed the case! From this perspective, life is rather easy for a compiler, which can do a pre-processing step and

• throw away any whitespace for good,
• remove all comments,
• expand all chains,
• apply all macros,
• …

and only then start doing its actual job (well, admittedly, there might be some complexity in that part…).

But not for ABAP cleaner! Because we care a lot about whitespace and alignment, have to stay aware that there may be any number of comments anywhere (including between any two tokens inside of a statement), must mind the pragmas (including in wrong places) and the chain colons …

So, unfortunately, so far, there was always a constellation that wasn't yet considered. But let's hope the new "stress test" which I mentioned in #147 helps to identify more of those in advance! :-)

And anyway, diligently writing regression tests, so at least we can be sure that whatever issue came up once for someone won't come up again for anyone! 🙂

Kind regards,
Jörg-Michael

[jmgrassau]

Hi Lars,

thanks again for reporting this – this issue should now be fixed in version 1.7.2, which was just released.

Kind regards,
Jörg-Michael