[Rosegarden-user] Fw: More info: Re: Bug report: unexpected "echo" while recording with external keyboard

[Jim C. <m_...@bu...>]

[I sent this yesterday, but I got an email back saying it had to be
manually accepted by a moderator because it was too large.  So I've
removed the two attachments in the original (midi file and error
output) to allow it to go through automatically.  If someone looks into
this and needs the originally attached files, let me know.]

Begin forwarded message:

Date: Sat, 29 Dec 2012 22:43:27 -0600
From: Jim Cochrane <m_...@bu...>
To: ros...@li...
Subject: More info: Re: [Rosegarden-user] Bug report: unexpected "echo"
while recording with external keyboard


On Tue, 25 Dec 2012 16:05:09 -0500
"D. Michael McIntyre" <ros...@gm...> wrote:

> On 12/22/2012 09:23 PM, Jim Cochrane wrote:
> 
> > This type of message showed up in the output 7 times when I
> > encountered the problem yesterday.
> 
> Those messages are usually meaningless.
> 
> I can't reproduce the problem, though I have no doubt you're really 
> experiencing it.  Wish I could be more helpful.

Thanks, Michael, for looking into this, and sorry for the late response.
It looks like one of those issues that are not only hard to reproduce,
but probably hard to track down the cause once it is reproduced.  (I'm
replying to the list because my response is probably of general
interest.)

I tried the following experiment, which may have duplicated this "echo"
issue:  I tried playing a MIDI file, which runs for about 12 to 15
minutes, with qtractor and sending its MIDI out to rosegarden's MIDI in
with rosegarden recording. This caused rosegarden to "sort of" freeze up
after several minutes.  (I say "sort of" because the yellow bar
continued to progress to the right, but the GUI was non-responsive -
e.g., hitting the "stop" key did nothing.)  Also, the same two warning
messages (as I reported in the original "echo bug" email) were output,
over and over again, when this problem occurred.

I don't expect you to look into this because, I suspect, the code that
handles MIDI recording is not your domain, and because I'm sure you
have plenty of other things you have to do.  But I'll report it here in
case someone on the RG team might find some time to look into it.  And,
perhaps, doing the same thing I did, with the same MIDI file, might
reproduce the problem for others.

If someone does look into this, I'm moving in the next couple days and
will not have internet access for a day or two, so if you send me
questions I might not respond until late in the week.  I'm in a bit of
a hurry, so I'll just include a description of the problem I recorded
when it happened - a few days ago -, which has more detail.  I wrote it
quickly, so hopefully it's coherent enough.  And I'll attach (assuming
SF's mailing system allows attachments - I think it does) the MIDI file
that I used.

Here's my write-up:

This file, played in qtractor: mshtst2.mid - with qtractor's midi output
going to rosegarden's MIDI in and rosegarden recording, caused
rosegarden to appear to freeze up (GUI non-responsive) and to continue
recording or seeming to record (the midi file had finished when I
looked so there was no real MIDI input going to rosegarden at that
point) - the yellowish bar continuing to progress to the right as if it
was still recording; it would not stop when I hit the stop or the
red/record button.  Also, rosegarden was outputting (probably on
stderr) warning messages to the terminal window it was run in - a long
stream of 2-line messages to the effect of:

WARNING: SegmentNotationHelper::makeNoteViable(): No valid split for
event of duration 986 at 259677 (split duration 1440), ignoring
remainder WARNING: This is probably a bug; fix required

This is the same 2-line message pattern that I reported in the "echo"
bug report from 2012-12-22:

WARNING: SegmentNotationHelper::makeNoteViable(): No valid split for
event of duration 1822 at 26880 (split duration 1920), ignoring
remainder WARNING: This is probably a bug; fix required

I don't know how long this 2-line message was repeated, but it was
enough to fill the scroll buffer of my terminal window.  (Output saved
to file error-output .)

I don't know if this bug is the same as the "echo" bug I reported on Dec
22, but it came about under similar circumstances - long stream (over 10
minutes) of MIDI input produced (originally) on a MIDI keyboard and the
same 2-line warning message pattern.  However, when I experienced the
"echo" problem while playing a MIDI keyboard (recording with
rosegarden), rosegarden didn't freeze.  It could be that if I had kept
playing it would have frozen in the same way that it did for this test.