|
| From: | Auto mailings of changes to Lily Issues via Testlilyissues-auto |
| Subject: | [Lilypond-auto] [LilyIssues-auto] [testlilyissues:issues] #5788 New French Beamimg Approach |
| Date: | Tue, 03 Mar 2020 15:50:44 -0000 |
Pushed to staging (with a somewhat exuberant commit message) as
commit cf714372e43703b6a76e2abcfacbda912f66d662
Author: Torsten Hämmerle <address@hidden>
Date: Tue Feb 18 22:22:42 2020 +0100
Issue 5788: New French Beamimg Approach Completely new approach to French beaming. This will automatically tackle all kinds of not-yet resolved positioning problems caused by the current French beaming implementation. ========== BASIC IDEA ========== Hypothesis: The *only* difference between standard and French beaming is that French "inner group" stems will not pass through all the beams. THAT'S ALL! Nothing else about it! The current approach to generally shorten French stems from the very beginning causes many follow-up positioning problems that have to be remedied later-on in many different places by neutralizing this deviation somehow. Tuplet numbers (w/o tuplet brackets) have already been dealt with, but many other problems still remain. GENERAL SOLUTION (in short) - Junk all exceptions and do not distinguish between French and standard beaming at all (quite radical, but extremely helpful) - That way we can be sure that all positionings will exactly match the standard beaming case - Only at the very end, when it comes to actually printing the stem, it has to be shortened by the appropriate amount. GENERAL SOLUTION (more detailed) - Junk all (!) existing French beaming special treatment, using standard stem lenghts throughout for all calculations so that everything exactly matches the standard beaming case automatically. - The standard beam length is extremely important for all the positioning calculations in several ways: * Special beam properties as all kinds of minimum lengths will only work correctly for standard length beams and there's no need at all for introducing more and more exceptions for French beams * Stem ends are crucial for placement of all kinds of grobs such as articulations, fingerings, text scripts and they *must* behave as if all the stems had standard length (property unchanged!) for vertical positioning of grobs aligned to the stem ends. * ONLY when it comes to actually printing the stem, French shortening becomes relevant. So *only* the ly:stem::pring function will actually need to know about the new french-correction property, using (standard) "length" property minus new "french-correction" property. ============= MODIFICATIONS ============= modified: Documentation/changes.tely Mention that French beaming will now exactly behave like standard beaming in all beam positionings and all articulation, fingering, etc. placements will now be identical to the standard beaming case. Show an example comparing standard/French beaming (w/o displaying the coding, though. It's just about demonstrating the (now) exactly matching positioning. modified: input/regression/beam-french.ly (existing) regression file. One beam positioning flaw (too far away from the noteheads) can already be seen in the current test cases, but I've added a few more examples containing exceptionally critical cases. modified: scm/define-grob-properties.scm Introduce new property "french-correction" modified: lily/include/stem.hh Additional Real parameter for set_stem_positions to pass over french_correction in order to be able to leave standard stem length untouched and set both property "length" and "french-correction". modified: lily/stem.cc - Stem::set_stem_positions * additional parameter Real fc for french_correction value * Set Stem property "length" as before, but now containing the unaltered (i.e. "unfrenched") stem length * if (and only if) fc is non-zero, set "french-correction" property of the Stem - Callback Stem::print * retrieve the (now unfrenched) stem-length property * retrieve the (new) french-correction property (0 if non-existent) * subtract french_correction from the standard stem_length (for printing the correctly shortened stem) - Add new property "french-correction" to the interface modified: lily/include/beam.hh - New struct Beam_stem_length containing stem_y_ (as before, but unaltered by French beaming) plus a french_correction_ containing the Frech beaming stem length delta. - Function Beam::calc_stem_y adaped * return Beam_stem_length rather than a mere Real stem_y * Boolean parameter "french" replaced by int value french_count for passing over the individual number of beam translations the stem has to be shortened by (if applicable; 0 for standard beaming or standard-length outer group French beams). modified: lily/beam.cc - Junk (now) obsolete French beaming special treatment - Initialize new struct Beam_stem_length - Adapt Beam::calc_stem_y * return new struct Beam_stem_length * replace bool french by int french_count * in case of non-zero french_count, calculate french_correction respecting beam_translation and feather_factor * pass over (now unaltered!) stem_y + id as before plus french_correction via new struc Beam_stem_length - Adapt Beam::set_stem_lengths * Determine value of french_count by intersecting left and right Slices of the stem's beaming property * Call Beam::calc_stem_y passing over french_count * Call Stem::set_stem_positions now with an additional parameter so that finally, both of the Stem properties "length" and new "french-correction" can be set. modified: lily/beam-quanting.cc - Junk (now) obsolete French beaming special treatment - Call of Beam::calc_stem_y adapted to return a structure rather than a single Real value - REMARK: French beaming is irrelevant (and even obstructive) for beam quanting (i.e. positioning) and so this feature is deliberately not being used and when calling calc_stem_y, french_count 0 is being passed over in any case to suppress unnecessary calculations. modified: lily/tuplet-number.cc Completely junk (now) obsolete French beaming special treatment, no exceptions are needed anymore. ========= IMPORTANT: ========= All hitherto unresolved placement issues (articulations, fingerings, etc.) caused by French beaming are automatically remedied by this completely new French beaming approach, as we are using standard beam lenghts throughout for all calculations, callbacks, positionings, minimum-length rules, and so forth... Cheers, Torsten Rename and add new regression file Patch Set 2 The following changes have been applied following the reviewers' comments: Rename stru Beam_stem_length -> Beam_stem_end (Han-Wen) Rename property french-correction -> stem-end-shorten (Han-Wen) Make property stem-end-shorten (was: french-correction) internal (DAK) Corrections to doc text (Werner) New regression file added directly comparing french beaming to standard beaming (Werner) On my behalf: Modified changes.tely LilyPond example to show more variations in less space so that it will not stick out into the margin in PDF version. Removed an unneccessary if, just multiplying by int 0 in this case. Full test successfully performed. Patch set 3: Rename stem-end-shorten -> french-beaming-stem-adjustment Patch set 4: missing semicolon at end of statement (ahem...)
[issues:#5788] New French Beamimg Approach
Status: Fixed
Labels: Fixed_2_21_0
Created: Mon Feb 24, 2020 12:39 AM UTC by Torsten Hämmerle
Last Updated: Tue Mar 03, 2020 03:38 PM UTC
Owner: Torsten Hämmerle
Attachments:
https://codereview.appspot.com/557500043
This will automatically tackle all kinds of not-yet resolved positioning problems caused by the current French beaming implementation.
As this is quite a radical and complete re-design of the LilyPond's current French beaming approach, I've decided to open up this issue because a general French beaming overhaul can not be associated with individual bug-related issues.
The only difference between standard and French beaming should be that French "inner group" stems will not pass through all the beams. That's all! It's as easy as this!
Unfortunately, all that Elaine Gould has to say about French beaming is: "don't do it!".
The current approach to generally shorten French stems from the very beginning causes many follow-up positioning problems that have to be remedied later-on in many different places by neutralizing this deviation somehow.
Tuplet numbers (w/o tuplet bracktes) have already been dealt with, but many other problems stil remain.
I'll attach an Old French Beaming Comparison PDF demonstrating a full-range test of all kinds of French beaming cases produced by the current implementation (2.19.84, but the output is identical to 2.20 and current 2.21). They are erroneous (i.e. deviating from standard beaming) in most of the cases - please have a look!
Standard and French beaming side-by-side so that all the deviations can easily be spotted.
Prerequisite: a new stem-interface property "french-correction" (pun intended).
After uploading my patch to Rietveld, I'll attach another PDF, showing how the new French beaming functionality will deal with all these cases.
Cheers,
Torsten
PS: New French Beaming Comparison PDF attached.
Sent from sourceforge.net because address@hidden is subscribed to https://sourceforge.net/p/testlilyissues/issues/
To unsubscribe from further messages, a project admin can change settings at https://sourceforge.net/p/testlilyissues/admin/issues/options. Or, if this is a mailing list, you can unsubscribe from the mailing list.
_______________________________________________ Testlilyissues-auto mailing list address@hidden https://lists.sourceforge.net/lists/listinfo/testlilyissues-auto
| [Prev in Thread] | Current Thread | [Next in Thread] |