Re: New format for autobeaming rules

[Neil Puttock]

2009/7/24 Carl Sorensen <address@hidden>:

>> type check also for settings?
>
> Settings needs a list? type check, and I haven't seen one available
> in c++.  It shouldn't segfault, because we do a pair check before we
> use it.

ly_is_list (which is a wrapper for scm_list_p) (but see below)

>
> I can't use a pair check for the argument, because '() is valid for
> settings.
>
> I could use pair_or_empty, but it doesn't exist

You could use ly_cheap_is_list, which does exactly this.

, and when I tried to
> add it to flower/include/guile-compatibility.hh, where all the rest of
> the types are defined, it gave me errors.

You'd need to add it to lily-guile.hh.  guile-compatibility.hh is for
compatibility with guile 1.6.

>> type checks?
>
> Put in for time_signature, rule_type.
>
> Can't do one for beam_type, because it needs to be pair-or-symbol,
> and I couldn't figure out how to add it.

The long way:

Add this to lily-guile.hh,

inline bool ly_is_symbol_or_pair (SCM x) { return scm_is_symbol (x) ||
scm_is_pair (x); }

and this to init_func_doc () in function-documentation.cc,

ly_add_type_predicate ((void*) &ly_is_symbol_or_pair, "symbol or pair");

then you can use LY_ASSERT_TYPE with the new predicate:

LY_ASSERT_TYPE (ly_is_symbol_or_pair, beam_type, 4);

The shorter way (and better since this is unlikely to be used
elsewhere) is to do this:

SCM_ASSERT_TYPE (scm_is_symbol (beam_type) || scm_is_pair (beam_type),
                   beam_type, SCM_ARG4, __FUNCTION__, "symbol or pair");

>> sort

> But I couldn't find the header file that defined SCM through git
> grep.  Do you know which file I need to include?

I think it's libguile.h, which is included in lily-guile.hh.

The following works for me:

beam-setting-scheme.cc:

#include "beam-settings.hh"
#include "context.hh"

beam-settings.hh:

(inside #define)

#include "lily-guile.hh"

>> decided not to restore original setting?
>
> I had decided to restore it in a different form.  1/8 beams are (6), which
> is equivalent to (3) in (3 . 4).  All shorter beams will be (1 1 1).

Ah, so it is.  I'll get the hang of the new settings eventually. :)

Regards,
Neil