Re: "Stress test" for the regression tests

[Reinhold Kainhofer]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am Donnerstag, 30. Juli 2009 00:47:07 schrieb Neil Puttock:
> 2009/7/29 Patrick McCarty <address@hidden>:
> > Some of the programming errors from the regression tests are pretty
> > bizarre, but I'll try taking a look at some of them eventually.  :-)
>
> To help you get started, here's a text file I've compiled with my
> thoughts on the remaining issues; some of them are easily fixed (e.g.,
> text interface errors in musicxml) 

How would you fix this? By default, the TimeSignature Grob does not have the 
text-interface, however, if you use the #ly:text-interface::print function as 
stencil, you need to set the text...


> MusicXML programming errors:
> 
> 42a-MultiVoice-TwoVoicesOnStaff-Lyrics.ly:    cyclic dependency: 
> calculation-in-
> progress encountered for #'X-extent (PaperColumn)

I have no idea what this warning is about...

> 46a-Barlines.ly:                              Improbable offset for stencil: 
> inf staff space (x2; bad
> barline string, type not implemented yet)

Ah, that's a typo, the bar type for short barlines should be "'" like for tick 
(although I suppose that is not entirely correct, but it's the best 
approximation that is available without including gregorian.ly and using 
\divisioMaior)

> MusicXML warnings (valid):
> 
> 23b-Tuplets-Styles.ly:                          Tuplet brackets of curved
> shape are not correctly implemented

Yes, I'm trying to use a slur as tuplet "bracket", but apparently the anchor 
points are not right...

> 33e-Spanners-OctaveShifts-InvalidSize.ly:       Invalid octave shift size
> found: 27. Using no shift.

Yes, this is intended. The test is supposed to check how applications behave 
with uncommon octave shifts...

> MusicXML warnings (invalid? CHECKME):
> 
> 12a-Clefs.ly:                                   unknown clef type `None'

It seems that lilypond does not provide any possibility to not show any clef 
at all... That's what would be needed here. The "None" comes simply from the 
MusicXML clef name in that case, which is taken verbatim for clefs where I 
didn't find a proper lilypond implementation.


> 13a-KeySignatures.ly:                           No ordering for key
> signature alterations (x2) (outlandish keys)

That's a general problem for all keys with more than 7 sharps/flats... E.g. try
\relative { \key bis \major bis1}


> 41g-PartNoId.ly:                                no music found in score

Yes, that's intended. The MusicXML file contains a part with no id assigned, so 
it will not be included in the score structure. 
However, there might be a better implementation, since that work only contains 
this one single part without an ID, one can deduce, that it should be the only 
part in the piece and include it, too.
This is not implemented in Musicxml2ly, though.

>45d-Repeats-Nested-Alternatives.ly:             More alternatives than
>repeats.  Junking excess alternatives

Hmm, that seems to be a bug in musicxml, as it doesn't correctly detect that 
there are three voltas...

>                                             Barcheck failed got 9 expect 10

That's a consequence of the warning above (the third alternative is ignored, 
which causes one missing measure)...


> 61g-Lyrics-NameNumber.ly:                     unterminated hyphen; removing

That's intended (the MusicXML file has a final lyrics syllable with 
<syllabic>begin</syllabic>).


> 61h-Lyrics-BeamsMelismata.ly:                   unterminated hyphen; 
removing

That's a general problem / inconsistency with lilypond and MusicXML: The 
MusicXML file has several slurred notes with lyrics attached, while lilypond 
does not allow assigning lyrics syllables to slurred notes... Thus the 
syllabled after the hyphen are ignored by musicxml.

> 99b-Lyrics-BeamsMelismata-IgnoreBeams.ly:     unterminated hyphen; removing

Same problem as 61h...

Cheers,
Reinhold

- -- 
- ------------------------------------------------------------------
Reinhold Kainhofer, address@hidden, http://reinhold.kainhofer.com/
 * Financial & Actuarial Math., Vienna Univ. of Technology, Austria
 * http://www.fam.tuwien.ac.at/, DVR: 0005886
 * LilyPond, Music typesetting, http://www.lilypond.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iD8DBQFKcX89TqjEwhXvPN0RAkAJAJ40TSfQrhenLjs6teZo36JeFTh8rgCfbgBc
HxOqiPlHwmXfdA2+NAxxZ6s=
=nLJZ
-----END PGP SIGNATURE-----