Transpiler WIP

[sbcgua]

No description provided.

[larshp]

@abaplint/transpiler-cli instead of @abaplint/transpiler

[larshp]

raise conv_failed.

raising non class based exceptions is not implemented, and I dont think it will be anytime soon

suggest refactoring to a class based exception

[sbcgua]

conv_failed

Hmm, yeah, I don't remember why I did like that in the first place

[sbcgua]

Nice, it compiles !

[sbcgua]

Looks like an operator is missing ?

    ro_parser.get().mt_components.set(await abap.Classes['ZCL_TEXT2TAB_UTILS'].describe_struct({i_struc: ro_parser.get().mo_struc_descr, i_is_deep: abap.builtin.boolc(abap.compare.initial(i_deep_provider) === false), i_ignore_nonflat: i_ignore_nonflat}));
    if ((abap.compare.initial(i_amount_format)abap.compare.initial(i_amount_format.getOffset({offset: 1, length: 1})))) {
      ro_parser.get().mv_amount_format.set(i_amount_format);
    }

[larshp]

yea, looks like a simple upper/lower case problem

[larshp]

a trick: you can set private: true so that it cannot accidently be published to npm

[sbcgua]

👍

[larshp]

next problem: Void type: RS38L_FNAM

Perhaps type as TFDIR-FUNCNAME instead, TFDIR will probably be added to open-abap sometime. RS38L_FNAM is a strange data element name

[sbcgua]

Actually funcname DE is a good replacement. Part of SABP pacakge

TFDIR refers RS38L_FNAM too in fact ...

[sbcgua]

@larshp Trying to do this again ... May I ask for some assistance ? The message says something wrong with /ui2/cl_json.clas which is not used anyhow ... :/

[larshp]

try updating these to latest

[larshp]

anyhow, next problem to tackle is VERI_ALPHA, not really sure what to use instead

ideally something that is marked as released, https://abapedia.org/steampunk-2305-api/

[sbcgua]

update didn't help: same error:

$node transpiled/index.mjs
file:///.../text2tab/transpiled/%23ui2%23cl_json.clas.locals.mjs:281
lcl_parser.ty_data_tt = abap.types.TableFactory.construct(new abap.types.Structure({"parent": new abap.types.String({qualifiedName: "LCL_PARSER=>TY_DATA-PARENT"}), "name": new abap.types.String({qualifiedName: "LCL_PARSER=>TY_DATA-NAME"}), "full_name": new abap.types.String({qualifiedName: "LCL_PARSER=>TY_DATA-FULL_NAME"}), "value": new abap.types.String({qualifiedName: "LCL_PARSER=>TY_DATA-VALUE"}), "type": new abap.types.String({qualifiedName: "LCL_PARSER=>TY_DATA-TYPE"})}, "lcl_parser=>ty_data"), {"withHeader":false,"keyType":"DEFAULT","primaryKey":{"name":"primary_key","type":"STANDARD","isUnique":false,"keyFields":[]},"secondary":[]}, "lcl_parser=>ty_data_tt");
                                                ^

TypeError: Cannot read properties of undefined (reading 'construct')
    at file:///.../text2tab/transpiled/%23ui2%23cl_json.clas.locals.mjs:281:49

... And how do you see from it, that it is about VERI_ALPHA ?

[sbcgua]

:/ I guess the only way is to create own DOMA ...

[larshp]

ALPHA: the xml generation might be bugged and remove the conversion routine

package.json: latest @abaplint/runtime is 2.7.54

[sbcgua]

Added domain/dtel to "deps" but no effect - same error

[larshp]

try updating package.json to latest versions of the dependencies

[sbcgua]

try updating package.json to latest versions of the dependencies

Yes, it's done 957b209

[sbcgua]

Ah no, sorry (need to get rid of the package json in favor of script, it just makes things worse).
But there is now another error

$node transpiled/index.mjs
Void type: FLOAT

[sbcgua]

OK, now it is some issue with conv_exit

ZCL_TEXT2TAB_PARSER: running ltcl_text2tab_parser_test->create
ZCL_TEXT2TAB_PARSER: running ltcl_text2tab_parser_test->apply_conv_exit
<ref *1> kernel_cx_assert [Error]
    at cl_abap_unit_assert.assert_equals (file:///C:/Users/at/Documents/devs/Abap/text2tab/transpiled/cl_abap_unit_assert.clas.mjs:355:32)
    at async ltcl_text2tab_parser_test.apply_conv_exit (file:///C:/Users/at/Documents/devs/Abap/text2tab/transpiled/zcl_text2tab_parser.clas.testclasses.mjs:312:5)
    at async run (file:///C:/Users/at/Documents/devs/Abap/text2tab/transpiled/index.mjs:45:9) {
  me: ABAPObject {
    qualifiedName: undefined,
    RTTIName: undefined,
    value: [Circular *1]
  },
  previous: ABAPObject {
    qualifiedName: 'CX_ROOT',
    RTTIName: '\\CLASS=CX_ROOT',
    value: undefined
  },
  textid: Character {
    constant: false,
    length: 32,
    extra: {},
    value: '                                '
  },
  get_longtext: [AsyncFunction: if_message$get_longtext],
  get_text: [AsyncFunction: if_message$get_text],
  actual: String { value: '123', qualifiedName: 'STRING' },
  expected: String { value: '0000000123', qualifiedName: 'STRING' },
  msg: String {
    value: "Expected '0000000123', got '123'",
    qualifiedName: 'STRING'
  },
  EXTRA_CX: {
    INTERNAL_FILENAME: 'cl_abap_unit_assert.clas.abap',
    INTERNAL_LINE: 349
  }
}

[larshp]

reproduced in unit test in open-abap/open-abap-core#720

[larshp]

try again

[larshp]

Void type: SYS_CALLS, suggest using type abap_callstack instead

[sbcgua]

Hmmm

  actual: String { value: '', qualifiedName: 'STRING' },
  expected: String { value: '', qualifiedName: 'STRING' },

[larshp]

the code calls fail(), so there is no actual and expected,

suggest adding the error code to the text in:

[larshp]

the classes are not in open-abap yet, and there is a nice but to fix in the transpiler to give a better error message in this scenario

[larshp]

abaplint/transpiler#1295

[larshp]

open-abap/open-abap-core@937cd4b part of open-abap/open-abap-core#737

[larshp]

btw, added abaplint/abaplint@b9c8242 to https://rules.abaplint.org/function_module_recommendations/

[larshp]

add these two skips,

      { "object": "ZCL_TEXT2TAB_PARSER", "class": "ltcl_text2tab_parser_test", "method": "parse_df16",  "note": "TypeError: abap.types.typeTodoDecFloat16Type is not a constructor" },
      { "object": "ZCL_TEXT2TAB_PARSER", "class": "ltcl_text2tab_parser_test", "method": "parse_ignore_convexit",  "note": "CUNIT issue" },

not feeling like implementing DecFloat16 this week..

[sbcgua]

:)

Done, though parse_ignore_convexit should theoretically work ... because the idea is to don't run the exit ... but can be checked later

[larshp]

current problem reproduced in open-abap/open-abap-core@2afe78a

[larshp]

two places where DMBTR is used,

[sbcgua]

fixed.

P.S. is there any way to make the error output less cryptic? I can potentially contribute ... though ... I'm not sure I understand the output enough. E.g. these look the same

  actual: String {
    value: 'name: MANDT, type: [object Object], as_include: , suffix: ',
    qualifiedName: 'STRING'
  },
  expected: String {
    value: 'name: MANDT, type: [object Object], as_include: , suffix: ',
    qualifiedName: 'STRING'
  },

[larshp]

fixed.

P.S. is there any way to make the error output less cryptic? I can potentially contribute ... though ... I'm not sure I understand the output enough. E.g. these look the same

  actual: String {
    value: 'name: MANDT, type: [object Object], as_include: , suffix: ',
    qualifiedName: 'STRING'
  },
  expected: String {
    value: 'name: MANDT, type: [object Object], as_include: , suffix: ',
    qualifiedName: 'STRING'
  },

for the current issue, the commented testcases are in open-abap/open-abap-core#739

problem is the RTTI in open-abap doesnt return identical object references for identical types

I implemented cl_abap_weak_reference earlier today, think I'm gonna make the RTTI classes serializable, instantiate everything, and compare with a weak reference cache, but not sure, its difficult

a bit less cryptic, but still cryptic message fixed in https://github.com/open-abap/open-abap-core/pull/739/files#diff-c761d101268df6f7b95ee1c9f81a84d0c9bbc2d15a262804315573b258e13439

[sbcgua]

Hi Lars, any progress with that issue? Can I do anything from my side?

[larshp]

not really sure what is the best way to fix the issue, perhaps just disable those testcases for now