Re: regtest bombs out with guilev2

[David Kastrup]

David Kastrup <address@hidden> writes:

> Thomas Morley <address@hidden> writes:
>
>> 2018-02-03 23:21 GMT+01:00 David Kastrup <address@hidden>:
>>
>>>
>>> Ok, then it's the primitive-eval crapping out.  parser.yy has
>>>
>>>         | markup_mode_word '=' identifier_init
>>>         {
>>>                 if (scm_is_false (Lily::markup_function_p ($3)))
>>>                 {
>>>                         parser->parser_error (@3, _ ("Not a markup 
>>> function"));
>>>                 } else {
>>>                         scm_primitive_eval
>>>                                 (scm_list_3
>>>                                  (Lily::define_markup_command,
>>>                                   scm_string_to_symbol ($1),
>>>                                   ly_quote_scm ($3)));
>>>                 }
>>>                 $$ = SCM_UNSPECIFIED;
>>>         }
>>>         ;
>>>
>>> So basically
>>
>> Compiling a file with nothing else then:
>>
>>> #(primitive-eval (list define-markup-command 'bold-red
>>>                        #{ \markup \bold \with-color #red \etc #}))
>>
>> returns:
>>
>> error: GUILE signaled an error for the expression beginning here
>> #
>>  (primitive-eval (list define-markup-command 'bold-red
>> source expression failed to match any pattern
>> fatal error: failed files: "atest-70.ly"
>
> Ugh.  Ok, I think I got this.  I'll split define-markup-command into a
> macro just calling an internal function, and I'll use the internal
> function inside of the parser.  Calling macros from C++ seems to be a
> bad idea.

Huh.  Not so simple: there are a few define-public called with
constructed names.  I'll have to see what kind of non-syntax equivalents
are available for those.  Will take a few days because I have to do
something else tomorrow urgently.

-- 
David Kastrup