bug#24560: 25.1; With `--font 7x13', wrong font is used for Unicode characters

[Lars Ingebrigtsen <larsi@gnus.org>]

Eli Zaretskii <eliz@gnu.org> writes:

>> Sounds like a bug in the autoconfig stuff.
>
> Yes.

The logic is a bit confused.

    if test "x${with_xft}" != "xno"; then
      EMACS_CHECK_MODULES([XFT], [xft >= 0.13.0], [], [HAVE_XFT=no])

[...]

    if test "$HAVE_XFT" != "yes"; then
       dnl For the "Does Emacs use" message at the end.
       HAVE_XFT=no
       HAVE_FREETYPE=no
    else
       dnl Strict linkers fail with
       dnl ftfont.o: undefined reference to symbol 'FT_New_Face'
       dnl if -lfreetype is not specified.
       dnl The following is needed to set FREETYPE_LIBS.
       EMACS_CHECK_MODULES([FREETYPE], [freetype2])

       test "$HAVE_FREETYPE" = "no" && AC_MSG_ERROR(libxft requires libfreetype)
    fi
  fi				  # $HAVE_CAIRO != yes

So if somebody says explicitly --without-xft, then HAVE_XFT won't be
"yes", and we set HAVE_FREETYPE to "no" without checking.  And harfbuzz
depends on HAVE_FREETYPE being "yes", so harfbuzz depends on xft being
available, too.

Is this correct behaviour?

In addition, I think the configure script should fail if you say
--with-harfbuzz and harfbuzz isn't available.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no