David Kastrup [Tue, 22 Mar 2016 23:05:16 +0000 (00:05 +0100)]
Issue 4811/1: Allow property paths as scalars and in assignments
They cannot be embedded LilyPond since #{ 2 . 2 #} already is valid
music.
Property paths must have at least two syntactic elements to be
recognized as such and must not start with a Scheme expression.
Technically, 3 . #'() counts.
Thomas Morley [Tue, 22 Mar 2016 10:32:53 +0000 (11:32 +0100)]
Issue 4808 Make chord-naming robust against unsetting majorSevenSymbol
This will likely be changed anyway as soon as rewriting
chord-names is seriously tackled.
At least it might be a reminder to take care of the possibility
a user wants to unset majorSevenSymbol.
Regtest amended.
James Lowe [Tue, 8 Mar 2016 12:09:50 +0000 (12:09 +0000)]
Doc: CG - Re-organize information about 'Patches'
Issue 4794
The information about how to
create, manage, upload and
review patches (as well as
the 'Patch cycle') is in many
different places making it
awkward (and time consuming)
for devs - especially those not
used to the process - to
locate the information.
Some of the information for
'using git' to create patches
does overlap with informeation
on how to submit patches. This
results in a lot of unnecessary
'jumping' about the
documentation.
This patch simply moves
sections - no new information
has been added - although some
of the @node and @*section names
have been changed to help with
re-organization.
David Kastrup [Tue, 15 Mar 2016 12:50:05 +0000 (13:50 +0100)]
Issue 4803: Amend convert-ly 2.17.6 rules for \tweak
Double application was problematic. Since grob properties start with
lowercase letters or X or Y while grob names start with A-W, it can be
reliably avoided to do the incorrect double conversion
David Kastrup [Sun, 13 Mar 2016 22:08:29 +0000 (23:08 +0100)]
Issue 4800: Require path separators in assignments/overrides
Historical assignments/overrides like
Variable #'x #'y = 7
or
\override Glissando.bound-details #'left #'Y = #3
without dots between the various path components do not match the syntax of
\revert and music functions or other uses. Disallow the non-separation
by '.' or ',' of such path components.
A key (previously a symbol) can be either a non-negative
integer (obeying the predicate index?) or a symbol. When used for
overrides/reverts, the first elements of a path may not actually be
numbers but they are flagged as an error by other means: since a full
override path can be the concatenation of several key lists, the leading
elements are not treated differently by the list predicates.
Strictly speaking this should not be the case for key-list-or-music? or
key-list-or-symbol? predicates, but making them behave differently seems
unwarranted. Arguably key-list-or-symbol? should possibly be
key-list-or-key? instead.
David Kastrup [Mon, 7 Mar 2016 11:13:06 +0000 (12:13 +0100)]
Issue 4798/2: Prepare override/revert to deal with numeric subkeys
Those are fairly cosmetic changes (partly comments) to prepare for the
possibility of using override/revert with numeric subkeys while
retaining the necessity of using a symbol as the main property in a
Context.Grob.property.subproperties... n-tuple since the organization of
an "nalist" (an alist stack containing reversible subproperty overrides)
relies on a few special key values: currently pairs and booleans are
specially interpreted while the current implementation requires an
eq?-comparable first index for efficiency reasons.
James Lowe [Thu, 25 Feb 2016 06:59:07 +0000 (06:59 +0000)]
Doc: CG - Updates for building the LP Docs using XeLaTex
Issue 4780
Updates to the CG for building
the LP Docs using XeLaTex -
instead of the default pdfTex -
where certain symbols (asterix,
pilcrow, section obelix and
diesis signs) are not printed
in the PDF output.
An @warning{} has been added
with explanation on how to
install texlive-xetex.
Note: This does not mean XeTeX
is a mandatory requirement,
unlike the main TexLive packages
already documented.
This patch also includes
additional small typos
and grammar fixes.
David Kastrup [Sun, 6 Mar 2016 19:45:02 +0000 (20:45 +0100)]
Issue 4786: fee.fi = 0 was interpreted as as fee = 0 for undefined fee
If fee had been defined previously, the behavior became more sensible,
treating fee as an alist with a value of 0 for key fi. This is now
always the case.
(Optionally) embed ly source files inside generated PDF
By using the new -dembed-source-code command line switch,
users may now include their LilyPond source code in the
generated PDF (using the pdfmark EMBED feature); although
the PDF is not modified in any visible way, embedded .ly
files should appear in viewers that support this feature
(other viewers should degrade gracefully).
This patch adds a new ly:source-files function (with
an optional parser arg) that allows to retrieve the list
of (non distribution-provided) ly files involved in
making the current score.
David Kastrup [Sat, 27 Feb 2016 10:41:08 +0000 (11:41 +0100)]
Issue 4718/1: In-chord tie in cue causes segfault
This extends transposition functionality from Music expressions to
Prob in general since Stream_event also needs transposition in the
context of quoted music. While at top level, Stream_event is
processed using the container-agnostic transpose_mutable, this does
not extend to the actual contents, and at least a Stream_event's
articulation list does not consist of Music expressions but again of
Stream_event elements.
James Lowe [Fri, 19 Feb 2016 11:54:25 +0000 (11:54 +0000)]
Doc: NR - changing-defaults.itely - various improvements
Issue 4761
A variety of nitpicks and
improvements reported by
Federico Bruni
Some spelling corrections
and typos fixed.
The section on \set vs.
\override has been
edited in an attmpt to
simplify the explanation
for the reader.
Fixed a couple of TODOs.
Added a @c that references
an open issue to improve
the example for pure/unpure
containers as the example no
longer illustrates what it
used to as of LP 2.17.x +