Re: Work on Issue 1320

[Marc Hohl]

Am 06.04.2012 23:04, schrieb Carl Sorensen:
> On 4/5/12 12:51 PM, "Marc Hohl"<address@hidden>  wrote:
>
> > Does it make sense to replace the definitions in bar-line.cc/span-bar.cc
> > with the scheme equivalents? If yes, I'd draw a patch and would include
> > one example for integrating user-defined bar lines with the new approach
> > as a snippet (once I get the dashed bar line right in place ...)
> It seems to me that it makes sense to put the stencil building functions
> in Guile instead of in c++.  No sense adding the extra interface layer for
> scheme calls from c++.
>
> I would suggest that you make custom-bar-print-alist and
> custom-bar-glyph-alist context properties, so they will be documented and
> available for modification.
>
> Whenever I see Scheme code that says something like
>
> (do ((
>      ...
>       (set!  foo bar)))
>
> I cringe a bit. This looks like a direct statement-by-statement
> translation from c++, rather than an implementation in native Scheme
> idioms.
Well, that's what I did, more or less. I saw a do-loop in Nicolas' code,
and I am pretty sure he knows *a lot more* scheme than I do, and
at least I am glad that the code works (well, most of it). It surely can be
improved ...

>    set! creates a side effect for the procedure, and the Scheme
> gurus like to avoid side effects as much as possible (at least they did 25
> years ago ... am I really that old?).  I haven't looked carefully at your
> code to write an alternative, but it's likely that this code can be
> rewritten either as a recursive function or as a map, fold, or apply
> function, all of which are more native Scheme idioms.
>
> Anyway, overall this looks like great work, and I think we ought to get it
> into the code base.
Sounds good - thanks for your kind words!

Regards,

Marc
> Thanks,
>
> Carl