Re: Issue 1294 - Lyric ties

[Patrick McCarty]

On Sat, Feb 12, 2011 at 9:51 AM, Phil Holmes <address@hidden> wrote:
> I'm checking some of the lyric-based issues in case they've been fixed in
> the latest release, and came back across
>
> http://code.google.com/p/lilypond/issues/detail?id=1294

[snip]

> {
>  \time 3/4
>  \relative c' { c2 e4 g2 e4 }
>  \addlyrics { gran- de_a- mi- go }
>  \addlyrics { pu- "ro y ho-" nes- to }
>  \addlyrics { pu- ro~y~ho- nes- to }
>  \addlyrics {
>   \override LyricText #'font-name = #"DejaVu Sans"
>   pu- ro~y~ho- nes- to
>  }
> }
>
> Now - there's one other interesting feature here.  If I delete the fourth
> line of the lyrics above (the one with the explicit font) and run Lily
> 2.13.49, saving the PS file, it has the following in the postscript header:
>
> %%DocumentMedia: a4 595.28 841.89 80 () ()
> %%DocumentSuppliedResources: font CenturySchL-Roma
> %%DocumentSuppliedResources: font Emmentaler-20
> %%DocumentSuppliedResources: font MS-Gothic
> %%EndComments
>
> If I run the same file in 2.12.3 I get:
>
> %%DocumentMedia: a4 595.28 841.89 80 () ()
> %%DocumentSuppliedResources: font CenturySchL-Roma
> %%DocumentSuppliedResources: font DejaVuSans
> %%DocumentSuppliedResources: font Emmentaler-20
> %%DocumentSuppliedResources: font feta-alphabet20
> %%EndComments
>
> which is interesting - 2.12 has found a font with the glyph in, and has
> automatically used it.  2.13 doesn't do this.
>
> Sorry - but this, for me, is back to looking like a regression.

Yes, looking back at the tracker, my focus was in finding the reason
why 2.13 was not loading fonts correctly.  I completely disregarded
the regression from 2.12.3, and as a result, I didn't really
investigate the behavior of 2.12.3...

We need to revisit this issue to figure out why it works in 2.12.3,
but not in 2.13.

However...  I'm almost certain that this is an issue with the build
system.  Do we really want to block 2.13 for a font-related build
system regression?  There are other build system regressions, notably
when we bumped Pango from 1.24.4 to 1.26.0.  Over time, I'm starting
to believe that some regressions were introduced between Guile 1.8.0
and 1.8.7, too.

I am willing to bump this back to a critical regression, since it *is*
a regression.  But the question is if it should block a stable release
or not.

Thanks,
Patrick