Rosegarden

> You're right, it's hard to describe, and it's totally broken.  Note
> lengths
> are coming out many times longer than the original in some cases, and it
> sounds like some of the double tap bass notes are getting skipped on
> export
> too.

>
> Any insights, Tom Breton?

I promise I will have a look at it tomorrow morning.  I only just read
this email now.

	Tom Breton (Tehom)

First, a brief summary: This seems to be an import bug.  Import may be
ignoring noteoffs that co-incide with noteons.

I had a look at the files.  And thank you dusthillresident, they were very
helpful.  Some data points:

(Old news) Importing thefastestthingalive_.mid into RG shows the problem -
some notes are held far too long.  Invariably the affected notes were (in
the original) followed by the same pitch, usually immediately.  Notes are
either held very long or completely unaffected.

Importing it into RG 11.11 shows exactly the same problem.  So it's not
new in 12.04

Exporting thefastestthingalive.rg here gives exactly the same result.

And this was the big clue for me: Playing it from the command line via
aplaymidi sounds fine.  No held notes, just clean chords, same as the
original.  So I'm sure this is on the import end.

I'm not terribly familiar with the import end, I only worked on export. 
But IIRC import does try to keep track of noteons and noteoffs and if
memory serves, does sometimes erase old ones.  Right now my theory is that
it somehow sees a noteon before it sees the noteoff for the previous note
of the same pitch, and so it loses the noteoff and thinks the note lasts
forever.

I haven't dived into fixing it yet.  I will when I finish this post.

> sounds like some of the double tap bass notes are getting skipped on
> export
> too.

The bass line looks and sounds the same to me.  I looked at the imported
segment just now and the ~moto perpetuo~ of eighth notes doesn't seem to
be missing any.  Could you double-check please, Michael?  Because if you
are seeing missed eighth notes, then we're not seeing the same thing.

	Tom Breton (Tehom)

On Saturday, June 09, 2012, Tom Breton (Tehom) wrote:

> I'm not terribly familiar with the import end, I only worked on export.

> I haven't dived into fixing it yet.  I will when I finish this post.

I really appreciate the detailed research.  I was thinking export, just 
because some of your most recent work had an impact on export, and because of 
the timing of the bug report.

Now that you've determined that it's an import bug, I think you're probably on 
the right track, and it's just that this particular composition happened to 
provide the key test pattern to reveal a much older problem.

If you can do something with that, I'll be grateful, but it's no longer a 
matter of asking you to please fix something you broke.  It doesn't look like 
you had anything to do with this one at all.
 
> The bass line looks and sounds the same to me.  I looked at the imported
> segment just now and the ~moto perpetuo~ of eighth notes doesn't seem to
> be missing any.  Could you double-check please, Michael?  Because if you
> are seeing missed eighth notes, then we're not seeing the same thing.

It'll have to wait until tomorrow or later, but I'll look back into it more 
closely.
-- 
D. Michael McIntyre

Looks like I spoke too soon.  My theory about what caused it seems to be
wrong, so now I'm still looking for the cause.  Doesn't seem to be on the
import end as I thought.

	Tom Breton (Tehom)

On Saturday, June 09, 2012, Tom Breton (Tehom) wrote:

> Looks like I spoke too soon.  My theory about what caused it seems to be
> wrong, so now I'm still looking for the cause.  Doesn't seem to be on the
> import end as I thought.

Easy to test this question.  Load his original file.  Export as MIDI.  Load 
that into MusE and witness the carnage with their equivalent of our matrix 
editor.  Lots of crazy long note bars in that piano chord part.

It probably plays OK because the programs used don't happen to be capable of 
sustaining indefinitely, so the notes eventually decay long before their note 
offs.
-- 
D. Michael McIntyre

Now I do have the solution.  And it's a boy-is-my-face-red one.

Two, actually.  I don't know why I thought that 11.11 importing it wrong
meant that 11.11 was exporting it wrong and I should look at importing.

The problem was that I was storing noteoffs in a set instead of a
multiset.  That would have worked except that I only compared noteoffs by
time, not by the whole noteoff.  Result: Two noteoffs couldn't exist at
the same instant in the same segment.  In importing, we hunt out as far as
it takes to find a matching noteoff, so the notes get longer and longer. 
This is what we were seeing.

I have coded a fix for it.  I will push it immediately.

	Tom Breton (Tehom)	T

On Saturday, June 09, 2012, Tom Breton (Tehom) wrote:

> I have coded a fix for it.  I will push it immediately.

And I'll try to knock together a 12.06 release this weekend to get this and a 
couple of other fixes out.
-- 
D. Michael McIntyre

D. Michael McIntyre <michael.mcintyre@...> writes:

> 
> On Saturday, June 09, 2012, Tom Breton (Tehom) wrote:
> 
> > I have coded a fix for it.  I will push it immediately.
> 
> And I'll try to knock together a 12.06 release this weekend to get this and a 
> couple of other fixes out.

I'm having exactly this problem with Rosegarden 12.04 in Debian wheezy
(installed from Debian repos).

On 10/22/2012 06:29 PM, tizo wrote:

>> And I'll try to knock together a 12.06 release this weekend to get this and a
>> couple of other fixes out.
>
> I'm having exactly this problem with Rosegarden 12.04 in Debian wheezy
> (installed from Debian repos).

I went on a search expedition through the archives to remind myself what 
this thread was about.

Thanks to you, now I remember the big important fix I was planning to 
base a quick bug fix release around a few months ago.

The release hasn't happened yet, partly because I was extremely busy all 
summer working 80+ hours every week, and partly because new things have 
continued to come along and get solved over the past few months.

The next release is going to be 12.12, it's due out in mid December. 
It's always a fine line between pushing the fixes sooner and releasing 
more often, and waiting for projects to wrap up to put out a more stable 
release less often.  The way this thing has been unfolding, it's 
favoring waiting on a more stable Rosegarden to emerge.

12.12 should be one of the most stable releases of Rosegarden to come 
along in quite some time, but it's not quite finished baking yet.
-- 
D. Michael McIntyre

> 
> 12.12 should be one of the most stable releases of Rosegarden to come 
> along in quite some time, but it's not quite finished baking yet.

Thanks for the quick response Michael. I think that Debian wheezy was 
frozen a couple of months ago. That's mean (afaik) that not new software 
versions will we added to the release. As wheezy is not yet the stable 
release, we could have several years of this Rosegarden bug in Debian. 
What is worse, is that Ubuntu could be affected too (as it is based on 
Debian).

So, maybe it would be a good idea to talk with the people from Debian, 
to patch the Rosegarden version 12.04 of their wheezy repo. I don't 
know other people, but I use this functionality extensively to pass 
new ideas to other musicians, as most of them don't use Rosegarden.

Just for the record, if there is someone with the same problem, you 
can workaround it by opening the Rosegarden file with an older version 
of the software, and export the file as a midi one. I could do that 
in an Ubuntu 12.04 with Rosegarden 11.something.

On 10/25/2012 02:39 AM, tizo wrote:

> release, we could have several years of this Rosegarden bug in Debian.

That's pretty much how the game works.  Most of the bugs we're working 
on now weren't reported until just recently, though they were introduced 
over a year ago. The release we're putting out now won't be in most 
people's hands until the end of 2013 or beyond, and those bugs will be 
fixed in the 2014 release, which will introduce new bugs that nobody 
will report until 2015, and so it goes.
-- 
D. Michael McIntyre