bug#35024: 26.1; unicode combining characters displayed incorrectly in emacs-26.1

[Mayer Goldberg]

Dear Eli, Basil:

I used your configuration features to indicate which libraries I need to install (added much more than just m17n and libotf), re-configured, and re-built, and now everything is fine. Thank you very much! I've learned a lot from this process. Thanks for all your help!

Mayer

On Thu, Mar 28, 2019 at 6:12 PM Eli Zaretskii <eliz@gnu.org> wrote:

> From: "Basil L. Contovounesios" <contovob@tcd.ie>
> Date: Thu, 28 Mar 2019 00:17:42 +0000
> Cc: 35024@debbugs.gnu.org
>
> > Hello:
> >
> > ?ש = \u05e9 = "HEBREW LETTER SHIN"
> > ?ָ = \u05b8 = "HEBREW POINT QAMATS"
> >
> > i.e., "\u05e9\u05b8", should display with the qamats right under the shin,
> > which is the case under emacs 25 and earlier, but fails in emacs 26: The
> > qamats is printed below but to the side, and not right under the letter
> > shin. This happens regardless of the font used, and with professional
> > OTF fonts that are designed to handle combining characters correctly,
> > and which are handled correctly by emacs 25 and earlier. This problem is
> > not unique to Hebrew, but is true for any RTL language I've checked
> > inclinding Arabic and Farsi.
>
> FWIW, I don't see the behaviour you describe from 'emacs -Q' using a
> checkout of the emacs-26 branch (what will be Emacs 26.2):

Right, and neither do I.

I think the reason why Mayer doesn't see correct shaping of
diacriticals is that his Emacs is built without libotf and m17n-flt:

> Configured features:
> XPM JPEG TIFF PNG RSVG SOUND GSETTINGS NOTIFY GNUTLS LIBXML2 FREETYPE
> XFT ZLIB TOOLKIT_SCROLL_BARS GTK2 X11 THREADS

Whereas yours, Basil, does have those libraries:

> Configured features:
> XAW3D XPM JPEG TIFF GIF PNG RSVG IMAGEMAGICK SOUND GPM DBUS GSETTINGS
> GLIB NOTIFY ACL LIBSELINUX GNUTLS LIBXML2 FREETYPE M17N_FLT LIBOTF XFT
> ZLIB TOOLKIT_SCROLL_BARS LUCID X11 XDBE XIM MODULES THREADS LIBSYSTEMD
> LCMS2

These two libraries are required for complex script shaping, see
README.  Crystal ball says Mayer's Emacs 25 was linked against those
two libraries, that's why it works there.