[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Issue 1586 in lilypond: midi cresc happens instantly

From: lilypond
Subject: Re: Issue 1586 in lilypond: midi cresc happens instantly
Date: Sun, 03 Apr 2011 09:00:47 +0000

        Status: Fixed

Comment #4 on issue 1586 by jan.nieuwenhuizen: midi cresc happens instantly

Fixed in 93b7a6f.

With the attached .ly it is easier to verify.

@hanwen: possibly.  I recently started a combined midi/midi2ly test suite
in input/regression/midi/*.  It appears, however, that no .pngs are produced
in out-test/; only in out-www/ and thus regressions are not tracked yet.

To run it, I do:

    make -C input/regression/midi out=www check

and look at


@hanwen, @graham:  The problem with this regression and the suggested
text-file-based approach is that I made some major changes to the MIDI
backend: even if we had such a simple regression test it would have said:
CHANGES.  Well, we knew that.

The MIDI output is expected to be different, because it had (and still has)
major problems.  Yes, `it works', but nobody cared for about 10 years and
it did some pretty stupid things.

Possibly it's better to have a stupid test that fails easily and must be
reset, than have nothing...  A real good midi backend test would be the
equivalent of png-diff for audio...  Possibly my roundtrip .MIDI, midi2ly
and png-diff is a useful approach.  It would have been possible to catch
this regression if midi2ly would produce dynamic markings in the -midi.ly
(ie: it will be possible ... as soon as ...) in the .ly.
Currently midi2ly ignores those...

        c.ly  121 bytes
        c.midi  189 bytes

reply via email to

[Prev in Thread] Current Thread [Next in Thread]