On 26 mars 2013, at 05:58, "Keith OHara" <k-ohara5...@oco.net> wrote:
> On Mon, 25 Mar 2013 00:29:35 -0700, m...@mikesolomon.org > <m...@mikesolomon.org> wrote: > >> On 25 mars 2013, at 07:10, k-ohara5...@oco.net wrote: >> >>> It looks like you try to use a common UP/DOWN direction for the portions >>> of a broken slur, and the image you posted to the bug-tracker showed a >>> common direction for each half of a broken slur, but the current patch >>> gives me inconsistent directions (in every case but especially line 3). >> >> I only use a common UP/DOWN if it is set by the user beforehand. Otherwise, >> it is calculated with the usual callbacks. > > I see. I thought you were using the same machinery that breaks slurs at > line-breaks, which uses a common direction between halves. I see now that > you are breaking earlier, at 'engraving', before all the notes under a slur > are engraved, thus befor the slur directions are determined. Shucks. I can look into this. > >>> Please do just one manual review of the regression suite between >>> versions before adding another test of this length. >> >> What do you mean here? > > You'll see a large number of very small regression tests, and a few huge very > repetitive tests. You might think the huge repetitive tests get in the way > of review. > I wanted to exhaust as many possibilities as I could think of (broken slurs over one note, several notes, beginning broken, and broken, etc.) for both slur and phrasing. Would it be better to split each system up into a separate test? Get rid of some tests? Cheers, MS _______________________________________________ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel