@unnumbered New features in 2.5 since 2.4
@itemize @bullet
+
+@item
+The PostScript backend is now used by default. This backend requires
+less machinery to run, and gives more consistent results.
+
+Due bugs in GhostScript's PDF handling, it is recommended to upgrade
+to GhostScript 8.x.
+
+@item
+Separator slashes may be inserted between systems in a score. See
+@file{input/regression/system-separator.ly} for an example.
+
+@item
+Locations of errors in the input are now calculated more precisely.
+
+@item
+LilyPond now uses Pango and FontConfig for selecting and rendering
+internationalized text in non-@TeX{} backends. A font may be selected
+by using a FontConfig name,
+
+@example
+\override TextScript #'font-name = #"Serif"
+@end example
+
+
+@noindent
+or using the classic font selection mechanism
+
+@example
+\override TextScript #'font-family = #'roman
+\override TextScript #'font-series = #'bold
+@end example
+
+@noindent
+Input strings encoded in UTF-8 are then rendered through Pango.
+
+@item
+Metrics of blocks of text can be retrieved from (La)@TeX{} directly,
+using the @code{-f texstr} output backend. This provides exact metrics
+for texts, including kerning and accents.
+
+@item
+LilyPond now uses FreeType read the Feta font as an OpenType
+font. This is cleaner design and more robust. A recent version of
+FontForge (11122004 or newer) and Freetype is required.
+
+As a result, the SVG backend is now a fully functional backend.
+
+@item
+A new script, @code{\espressivo} has been added, for a combination of
+crescendo and decrescendo on a single note.
+
@item
In markups, expressions stacked with @code{\column},
@code{\center-align}, etc, are not grouped with @code{< ... >} anymore,
but with @code{@{ ... @}}, eg:
@example
-\markup \column @{ @{ first line @} @{ second line @} @}
+\markup \column @{
+ \line @{ first line @}
+ \line @{ second line @}
+@}
@end example
@item