|
From: | Phil Holmes |
Subject: | Re: Lilypond benchmarking |
Date: | Sat, 21 Sep 2013 16:53:27 +0100 |
To: "David Kastrup" <address@hidden> Cc: <address@hidden> Sent: Saturday, September 21, 2013 4:37 PM Subject: Re: Lilypond benchmarking
----- Original Message ----- From: "David Kastrup" <address@hidden>To: "Phil Holmes" <address@hidden> Cc: "James" <address@hidden>; <address@hidden> Sent: Saturday, September 21, 2013 2:12 PM Subject: Re: Lilypond benchmarking"Phil Holmes" <address@hidden> writes:2.12 - 162 pages; 2.14 - 147; 2.16 - 142; 2.17.26 - 158pp. 2.17 is noticeably looser, but I concluded I'd adjust some of the spacing controls to fit more to a page.That's actually a real problem. Now 2.17.27 will have some padding significantly reduced (halved?) if I understand Keith correctly, so we should get some pages more out. If I remember correctly, the skyline code made it desirable to increase some paddings to avoid jamming things too closely. And also if I remember correctly, the page break decisions do _not_ make use of skylines. Instead, skylines are only used for spreading out the page _after_ pagebreaking, so the net result will be more pages rather than fewer. I may have understood something wrong here. However, if my understanding is _not_ mistaken here, I think we should not release 2.18 before we have integrated the interstaff positioning using skylines into the page breaking decisions. And of course, the distance between two successive skylines should be measured _once_ at most. It probably needs to become a part of vertical spacing rods or something. -- David KastrupI've done quite a bit of work trying to see what's going on and have discovered that something introduced between (I think) 2.17.18 and 2.17.19 has affected how lily determines whether it can fit another system in. With ragged-bottom set and annotate spacing, on one of my scores 17.19 shows 49.07 space left and the next system having an extent of 44.7. Despite that it puts that system on the following page. The earlier version fits it in, albeit with the complaint "warning: cannot fit music on page: ragged-spacing was requested, but page was compressed". Without ragged-bottom, that complaint is not there.Bisecting. -- Phil Holmes
51560f756aa3ab37592c815062e733998accf79c is the first bad commit commit 51560f756aa3ab37592c815062e733998accf79c Author: Keith OHara <address@hidden> Date: Wed May 25 23:43:08 2011 -0700 align-interface.cc: Clarify code for empty staves --Phil Holmes
[Prev in Thread] | Current Thread | [Next in Thread] |