<?xml version="1.0"?>
<rss version="2.0.">
  <channel>
    <title>Advogato blog for ensonic</title>
    <link>http://www.advogato.org/person/ensonic/</link>
    <description>Advogato blog for ensonic</description>
    <language>en-us</language>
    <generator>mod_virgule</generator>
    <pubDate>Wed, 14 May 2008 11:01:18 GMT</pubDate>
    <item>
      <pubDate>Sat, 10 May 2008 12:18:10 GMT</pubDate>
      <title>10 May 2008</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=79</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=79</guid>
      <description>&lt;b&gt;buzztard&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; As can be seen on our &lt;a href="http://www.buzztard.org/index.php/Roadmap" &gt;roadmap&lt;/a&gt;&#xD;
sample support is scheduled for 0.4. As this&#xD;
is the major feature, we started working on it. I have to&#xD;
say that building&#xD;
buzztard on top of GStreamer was definitely the right thing&#xD;
to do. We can now&#xD;
load whatever GStreamer can handle. FSM did a nice&#xD;
cairo-based waveform widget.&#xD;
Right now I am working on the code that allows plugins to&#xD;
access the wavetable.&#xD;
It will hopefully be ready real soon.&#xD;
&#xD;
&lt;p&gt; Besides that I've started to restructure the gst-buzztard&#xD;
package a bit further.&#xD;
I also plan to merge the buzz wrapper plugin into it. The&#xD;
aim is to reduce the&#xD;
number of packages that one needs to build.&#xD;
</description>
    </item>
    <item>
      <pubDate>Sat, 12 Apr 2008 20:05:21 GMT</pubDate>
      <title>12 Apr 2008</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=78</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=78</guid>
      <description>After the release I demoed it some friends and obviously&#xD;
found some issues. Now I&#xD;
spend the rest of the month tracking then. Its mostly&#xD;
ref-counts issues. Why are&#xD;
the so nasty? Well unlike memory leaks, they are anonymous.&#xD;
Memory is allocated&#xD;
once and should be paired by one free. The id of the&#xD;
resource is the memory&#xD;
address. Now every ref should be paired by an unref, but all&#xD;
the refs and unrefs&#xD;
share the same id (address of the object). So how to figure&#xD;
which ref is not&#xD;
having the unref? &lt;a href="http://refdbg.sf.net" &gt;Refdbg&lt;/a&gt;&#xD;
can trace object and gather backtraces for&#xD;
ref-count operations. So one has to go through the list one&#xD;
by one, remove pairs&#xD;
and study whats left. Its not easy, but it works. &#xD;
Now why is this so important. If the refcount never reaches&#xD;
0, the object is not&#xD;
disposed and occupies memory (until the application exits).&#xD;
Under valgrind it&#xD;
would appear under reachable memory usually. For most apps&#xD;
it has no immediately&#xD;
visible effect. For gstreamer app it has. In buzztard I do&#xD;
dynamic stuff with&#xD;
pipelines, like adding and removing elements. If you get the&#xD;
ref-counting wrong.&#xD;
The element is unlinked but not disposed. If the pipeline is&#xD;
set to PLAYING one&#xD;
gets data flow errors. But then its not a gstreamer only&#xD;
problem. E.g. icon theme&#xD;
pixbufs in gtk. If the are not unrefed, gtk cannot unload&#xD;
the theme. Means if&#xD;
you try a bunch of themes, none of the is unloaded if even&#xD;
just one panel&#xD;
applet has a ref-count issue. The sad thing is that the&#xD;
gnome libraries have&#xD;
quite some ref-count leaks. I've started to track and fix&#xD;
them as its hard to&#xD;
see something within all this noise&#xD;
(&lt;a href="http://bugzilla.gnome.org/show_bug.cgi?id=525815" &gt;#525815&lt;/a&gt;,&#xD;
&lt;a href="http://bugzilla.gnome.org/show_bug.cgi?id=66513" &gt;#66513&lt;/a&gt;).&#xD;
&#xD;
&lt;p&gt; I've released buzztard-0.3.1 with the fixes. A few more&#xD;
developments to mention:&#xD;
Waffel started a demo for the zip loader based on libgsf,&#xD;
Deloun made great svg&#xD;
designs for the machine view canvas and the tabs and Herzi&#xD;
send me a patch to&#xD;
add native scrolling in the pattern editor. Now back to new&#xD;
features, yah!</description>
    </item>
    <item>
      <pubDate>Thu, 27 Mar 2008 18:12:15 GMT</pubDate>
      <title>27 Mar 2008</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=77</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=77</guid>
      <description>&lt;img align="left" src="http://www.buzztard.org/files/buzztard_0.3_banner_88_31.png"&gt;The&#xD;
buzztard team has released version 0.3.0 "a tale of ice and&#xD;
darkness" of its buzz-alike music composer. All modules got&#xD;
extensive improvements over the last release from almost a&#xD;
year ago. It is usable now and fun to play with. Give it a&#xD;
try and &lt;a href="http://sourceforge.net/tracker/?group_id=55124&amp;atid=476037" &gt;report&#xD;
bugs&lt;/a&gt;.&#xD;
&#xD;
&lt;p&gt; &lt;b&gt;bml&lt;/b&gt; : Support for native machines. Better emulation.&#xD;
Can be build on 64bit x86.&#xD;
&#xD;
&lt;p&gt; &lt;b&gt;bsl&lt;/b&gt; : Support for pre 1.2 buzz songs. Handle volume and&#xD;
panorama&#xD;
on wires. Several bug fixes.&#xD;
&#xD;
&lt;p&gt; &lt;b&gt;buzztard&lt;/b&gt; : Improvements in all areas. We now have native&#xD;
buzzmachine support and better compatibility. The buzztard&#xD;
editor got an own pattern editor widget. The ui got&#xD;
lots of keyboard commands. We have settings for default&#xD;
directories used. The ui has dialogs for recording mix-downs&#xD;
and also single tracks.&#xD;
&#xD;
&lt;p&gt; &lt;b&gt;gst-buzztard&lt;/b&gt; : Improvements on the preset interface.&#xD;
Support for sparse streams (GAP flag).&#xD;
&#xD;
&lt;p&gt; &lt;b&gt;gstbml&lt;/b&gt; : Sparse stream handling. Better&#xD;
compatibility. Bug fixes.&#xD;
&#xD;
&lt;p&gt; &lt;a href="http://www.buzztard.org" &gt;project-page&lt;/a&gt; -&#xD;
&lt;a href="http://www.buzztard.org/index.php/Screenshots" &gt;screenshots&lt;/a&gt;&#xD;
-&#xD;
&lt;a href="http://sourceforge.net/project/showfiles.php?group_id=55124" &gt;downloads&lt;/a&gt;&lt;br&gt;&#xD;
</description>
    </item>
    <item>
      <pubDate>Thu, 6 Mar 2008 19:30:33 GMT</pubDate>
      <title>6 Mar 2008</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=76</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=76</guid>
      <description>&lt;b&gt;buzztard&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; The &lt;a href="http://lac.linuxaudio.org/" &gt;Linux Audio&#xD;
Conference 2008&lt;/a&gt; was nice once again. Good talks,&#xD;
interesting know-how exchange and of course opportunities&#xD;
for testing and hacking on buzztard.&#xD;
&#xD;
&lt;p&gt; I have improved the playback engine a bit. It now uses less&#xD;
additional latency.&#xD;
Still one get much lower latency by using jack, instead of&#xD;
pulse audio (currently experimenting to lower latency there&#xD;
too).&#xD;
&#xD;
&lt;p&gt; The pattern editor got better gtk+ theme support (especially&#xD;
on dark ones) and&#xD;
also tries to use the default monospace font and font-size&#xD;
to better match with&#xD;
the overall look of your desktop. I implemented blending&#xD;
value ranges and&#xD;
randomization of ranges. FSM implemented more keyboard&#xD;
shortcuts. I added value&#xD;
descriptions to the status bar where the plugin supports it.&#xD;
&#xD;
&lt;p&gt; Also the sequence editor got some todo-fixing. The sequence&#xD;
was never&#xD;
horizontally scrollable (adding columns would force the&#xD;
window to be wider). The&#xD;
M,S,B buttons are now color shaded when pressed to give&#xD;
better visual clues.&#xD;
Finally also the position unit switcher is functional&#xD;
(ticks/time) and the label&#xD;
navigator gets properly updated. Lots of small things I&#xD;
never got around until&#xD;
now&#xD;
&#xD;
&lt;p&gt; Level meters in sequence and analyzer windows are now more&#xD;
lively and precise.&#xD;
Having several analyzers open now also produces the expected&#xD;
results.&#xD;
&#xD;
&lt;p&gt; The buzz song importer got further fixes. It seems to load&#xD;
all of my songs just&#xD;
fine now.&#xD;
&#xD;
&lt;p&gt; Some improvements also on the buzz emulation side. More&#xD;
machines load and work.&#xD;
We have a nice testsuite with html reports now. I spotted&#xD;
some issues which will&#xD;
fix more machines. Will try to implement this next.&#xD;
&#xD;
&lt;p&gt; If nothing urgent comes up. I'll prepare the release in the&#xD;
coming days. Stay&#xD;
tuned.&#xD;
</description>
    </item>
    <item>
      <pubDate>Thu, 7 Feb 2008 20:42:50 GMT</pubDate>
      <title>7 Feb 2008</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=75</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=75</guid>
      <description>&lt;b&gt;buzztard&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; We're heading for a release. There is a few more features&#xD;
that we'd like to&#xD;
finalize. Then its time for testing and this is were *you*&#xD;
can help.&#xD;
Featurewise buzztard improved nicely this month. I improved&#xD;
stereo support.&#xD;
We now have new look for the wires. They show animated&#xD;
volume levels and&#xD;
panorama positions. Both volume and panorama can be adjusted&#xD;
via popup-sliders.I&#xD;
like this a lot more that how it works in buzz. There it was&#xD;
a bit confusing in&#xD;
some cases (not clearly indicating which connections a&#xD;
panorama setting would&#xD;
affect). As an extra bonus, buzztard does stereo balance on&#xD;
stereo connections.&#xD;
&#xD;
&lt;p&gt; I also updated the buzz song importer to make use and&#xD;
correctly import panorama&#xD;
and volume. Now also the master-volume is mapped to the&#xD;
scale used in buzztard.&#xD;
&#xD;
&lt;p&gt; FSM worked on the pattern widget. It supports variable&#xD;
cursor stepping (going&#xD;
down several lines at once) and selection. I added&#xD;
column/group/all insert and&#xD;
delete actions.&#xD;
&#xD;
&lt;p&gt; Finally Waffel wrote a &lt;a href="http://www.buzztard.org/index.php/Building#Building_with_Jhbuild" &gt;jhbuild&#xD;
module set for buzztard&lt;/a&gt;.&#xD;
</description>
    </item>
    <item>
      <pubDate>Tue, 8 Jan 2008 20:21:03 GMT</pubDate>
      <title>8 Jan 2008</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=74</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=74</guid>
      <description>&lt;b&gt;buzztard&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; I guess tracker-style music editors have to live with the&#xD;
heritage of tabular&#xD;
pattern input. Some feature rely heavily on it. FSM started&#xD;
with such a widget&#xD;
and I integrated it into buzztard. One can still switch&#xD;
between the previous and&#xD;
the new one by setting a define. The widget works quite nice&#xD;
already. The only&#xD;
missing feature is native scrolling support. To add that to&#xD;
an own widget&#xD;
requires an ugly hack. I can't figure why gtk does not use&#xD;
an interface for it.&#xD;
&#xD;
&lt;p&gt; Patterns in buzz also carry volume and panorama events for&#xD;
the inputs of the&#xD;
machine. This was totally missing in buzztard. Now we have&#xD;
quite nice support&#xD;
for it. The volume control works and panorama/balance&#xD;
support is next on the&#xD;
to-do list. While working on that I made master-machine&#xD;
patterns useful by&#xD;
allowing to control master-volume,&#xD;
&#xD;
&lt;p&gt; The sequence view got more commands. Finally one can insert&#xD;
and delete rows in&#xD;
all columns or only in the current track.&#xD;
&#xD;
&lt;p&gt; If that wasn't enough, we also did a round of maintenance.&#xD;
All code has been&#xD;
moved to svn. The buildbot has been extended to build all&#xD;
modules now.&#xD;
And finally I got around to port the code to not use the&#xD;
deprecated gtk tooltip&#xD;
API if new gtk is available.&#xD;
&#xD;
&lt;p&gt; That's a great start for 2008 I'd say!</description>
    </item>
    <item>
      <pubDate>Mon, 3 Dec 2007 19:17:47 GMT</pubDate>
      <title>3 Dec 2007</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=73</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=73</guid>
      <description>&lt;b&gt;buzztard&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; After porting some simpler machines last month, FSM and&#xD;
myself where looking at&#xD;
infector. This already is a quite complex generator. Porting&#xD;
as such wasn't that&#xD;
hard, but unfortunately it did not play. But with the&#xD;
sources at hand we could&#xD;
figure out what the issues was and fix the problem. It was a&#xD;
generic problem in&#xD;
the buzzmachine host emulation. This could potentially also&#xD;
make other machines&#xD;
work. Note to myself - I need to re-test the blacklisted ones.&#xD;
One outcome of the better machine compatibility is that some&#xD;
of the original&#xD;
buzz song now plays with some extra manual help.&#xD;
&#xD;
&lt;p&gt; Some improvements on the UI side also. The spectrum analyzer&#xD;
window is now&#xD;
re-sizable. We also have a workaround for a gtk+ bug &lt;a&#xD;
href="http://bugzilla.gnome.org/show_bug.cgi?id=131177"&gt;[1]&lt;/a&gt;,&#xD;
there is now an&#xD;
accelerator to start playback.&#xD;
&#xD;
&lt;p&gt; Unfortunately my recent attempts to merge my preset iface &lt;a&#xD;
href="http://bugzilla.gnome.org/show_bug.cgi?id=396779"&gt;[2]&lt;/a&gt;&#xD;
to gstreamer got &#xD;
stopped again. I need to try harder poking the other&#xD;
developers to review it.&#xD;
&#xD;
&lt;p&gt; That's it for this month - happy x-mas month to everybody.&#xD;
&#xD;
</description>
    </item>
    <item>
      <pubDate>Sat, 10 Nov 2007 18:55:54 GMT</pubDate>
      <title>10 Nov 2007</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=72</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=72</guid>
      <description>&lt;b&gt;buzztard&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; Very good news for buzz lovers. Recently FSM joined&#xD;
#buzztard irc channel. He&#xD;
gave us access to his buzzmachine sources and the okay to&#xD;
publish them under&#xD;
LGPL. One thing that I had on my list for too long was to&#xD;
extend my&#xD;
buzz-emulation layer to support natively built buzzmachines.&#xD;
Well now its in CVS,&#xD;
A bml/gstbml rewrite to support native machines + initial 5&#xD;
machines working.&#xD;
Means approximately 900 to go. One thing that we need to be&#xD;
able to port some&#xD;
machines is a native implementation of buzz's dsplib. We&#xD;
will look into that&#xD;
next and of course port more machines.&#xD;
&#xD;
&lt;p&gt; Some small progress also in the UI. I've added track&#xD;
reordering in sequence tab&#xD;
and keymap independent note entering (y vs z mapping on&#xD;
different key layouts),&#xD;
</description>
    </item>
    <item>
      <pubDate>Sat, 20 Oct 2007 19:17:53 GMT</pubDate>
      <title>20 Oct 2007</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=71</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=71</guid>
      <description>&lt;b&gt;opensuse 10.3&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; Its the time of the year when I ruin my setup by upgrading&#xD;
my opensuse box. The actual upgrade went surprisingly&#xD;
smooth. The only hickup was that all disk are now named sd*&#xD;
instead of hd* and the setup was not prepared to update&#xD;
grub. With some manual intervention on a console and&#xD;
changing the settings by hand I could proceed.&#xD;
&#xD;
&lt;p&gt; The reboot worked fine and it thought for a moment that&#xD;
everything is fine. Its not:&lt;ul&gt;&#xD;
&lt;li&gt;&lt;a&#xD;
href="https://bugzilla.novell.com/show_bug.cgi?id=333999"&gt;333999&lt;/a&gt;&#xD;
gnome-control-center missing in the menu&#xD;
&#xD;
&lt;li&gt;&lt;a&#xD;
href="https://bugzilla.novell.com/show_bug.cgi?id=334050"&gt;334050&lt;/a&gt;&#xD;
gnome window decorator does not show the application icon&#xD;
&#xD;
&lt;li&gt;&lt;a&#xD;
href="https://bugzilla.novell.com/show_bug.cgi?id=334062"&gt;334062&lt;/a&gt;&#xD;
gnome-xgl-switch and white desktop&#xD;
&amp;lt;/a&amp;gt;&#xD;
&lt;li&gt;&lt;a&#xD;
href="https://bugzilla.novell.com/show_bug.cgi?id=334081"&gt;334081&lt;/a&gt;&#xD;
totem has support for plugins, but they are missing in the&#xD;
package&#xD;
&#xD;
&lt;li&gt;&lt;a&#xD;
href="https://bugzilla.novell.com/show_bug.cgi?id=335267"&gt;335267&lt;/a&gt;&#xD;
/etc/init.d/kbd: setfont: putfont: 256,8x16: failed: -1&#xD;
&#xD;
&lt;li&gt;&lt;a&#xD;
href="https://bugzilla.novell.com/show_bug.cgi?id=335491"&gt;335491&lt;/a&gt;&#xD;
gnome-session hangs on logout&#xD;
&#xD;
&lt;/ul&gt;&#xD;
&#xD;
&lt;p&gt; I also filed some for the gnome components:&lt;ul&gt;&#xD;
&lt;li&gt;&lt;a&#xD;
href="http://bugzilla.gnome.org/show_bug.cgi?id=486873"&gt;486873&lt;/a&gt;&#xD;
Evolution: Switching the shown category in contact view&#xD;
kills or hangs evolution-data-server&#xD;
&#xD;
&lt;li&gt;&lt;a&#xD;
href="http://bugzilla.gnome.org/show_bug.cgi?id=488663"&gt;488663&lt;/a&gt;&#xD;
gnome-panel: gtk_recent_manager_add_full: assertion `uri !=&#xD;
NULL' failed&#xD;
&#xD;
&lt;li&gt;&lt;a&#xD;
href="http://bugzilla.gnome.org/show_bug.cgi?id=488710"&gt;488710&lt;/a&gt;&#xD;
gmenu-simple-editor: errors in .xsession-errors&#xD;
&#xD;
&lt;/ul&gt;&#xD;
&#xD;
&lt;p&gt; Some reports got comments already and and some got info from&#xD;
my side of how to fix it, for some I am still in the dark.&#xD;
Lets hope they get addressed quickly and fixes are made&#xD;
available via online-update. The evolution bug worries me, a&#xD;
bug report with a backtrace for a double free, that is&#xD;
low-hanging fruit, but no one commented yet. Regarding the&#xD;
gnome-panel one, there is quite a lot of stuff in the&#xD;
.xsession-errors. Is that simply the stderr of everything&#xD;
started from the gnome-session? Shouldn't we require apps to&#xD;
somehow prefix their output with a domain (or use g_log&#xD;
methods instead of fprintf).&#xD;
&#xD;
</description>
    </item>
    <item>
      <pubDate>Wed, 17 Oct 2007 15:47:25 GMT</pubDate>
      <title>17 Oct 2007</title>
      <link>http://www.advogato.org/person/ensonic/diary.html?start=70</link>
      <guid>http://www.advogato.org/person/ensonic/diary.html?start=70</guid>
      <description>&lt;b&gt;gstreamer debugging support&lt;/b&gt;&#xD;
&#xD;
&lt;p&gt; GStreamer pipelines can quickly become quite complex. Even a&#xD;
simple video playback pipeline easily plugs 20 elements. Now&#xD;
ts something goes wrong, developers had to go through&#xD;
zillion lines of log-output. Now there is some help:&#xD;
&#xD;
&lt;p&gt; &lt;a&#xD;
href="http://www.buzztard.org/files/gst-launch.0:00:00.984100000.png"&#xD;
border="0" align="center"&gt;&lt;img&#xD;
src="http://www.buzztard.org/files/gst-launch.0:00:00.984100000.small.png"&gt;&lt;/a&gt;&#xD;
&#xD;
&lt;p&gt; Today I committed the code for &lt;a&#xD;
href="http://bugzilla.gnome.org/show_bug.cgi?id=456573"&gt;#456573&lt;/a&gt;.&#xD;
All it needs is a simple macro in the code. The graph&#xD;
dumping gets disabled for non-debug builds. When running use&#xD;
GST_DEBUG_DUMP_DOT_FILES=1 in the environment to activate&#xD;
it. It dumps several graph by using timestamps in the&#xD;
filename. To get e.g. the png from the graph one runs:&#xD;
&#xD;
&lt;p&gt; dot -Tpng /tmp/gst-launch.0:00:00.984100000.dot&#xD;
-o/tmp/gst-launch.0:00:00.984100000.png&#xD;
&#xD;
&lt;p&gt; If now apps like totem, rhythmbox and co. put this into&#xD;
the handler for warning and error bus-messages, we have a&#xD;
nice way to track down playback problems.&#xD;
</description>
    </item>
  </channel>
</rss>
