You are here

Qtractor 0.4.6 - The Funky Deviless is here!

So much to tell, even more to do... then one could hardly shake, this long overdue. Lousy rhymes and no miserly times. And there it is: a bug-fix release, I'll mean to ease.

Oh crap! Let's get it through once and for all.

With huge compliments to all who got the nerve and report as many too much idiosyncrasies (nee bugs). Don't, never look back. There's plenty more ahead, no matter where you look, or hear, whether is up or down hill :)

Qtractor 0.4.6 (funky deviless) is here!

Release highlights:

  • MIDI Editor draw mode (aka paint mode) (NEW)
  • MIDI Swing-quantize (NEW)
  • LV2 UI Instance & Data-access extension support (NEW)
  • JACK Session support (EXPERIMENTAL) (NEW)
  • LV2 Save/Restore extension support (NEW)
  • MIDI Editor event list in-line editing (NEW)
  • MIDI Clip time-stretching (FIX)
  • MIDI Clip editor file salvage quietness (NEW)
  • MIDI Control bus switching crash (FIX)
  • MIDI Bank-selection backout (FIX)
  • Initial widget geometry extents (FIX)
  • Input-only bus playback crash (FIX)
  • Bus connection persistence crash (FIX)
  • Drag-and-drop cloning plugins (FIX)
  • MIDI Editor floating-selection persistence (NEW)
  • Audio inserts garbage signal (FIX)

A bit more or not so detailed change-log is found below.


Project page:


Weblog (upstream support; you're already reading it):

License (no kiddin'):

Qtractor is free, open-source software, distributed under the terms of the GNU General Public License (GPL) version 2 or later.


  • Introducing a non-painting edit sub-mode on the MIDI clip editor's piano-roll (see Edit/Select Mode/Edit Draw menu).
  • The MIDI clip editor (aka piano-roll) is now a lot more quiet about saving its own dirty content, delegating all
    salvage questions to main session control.
  • Don't show session restart message box when changing JACK transport mode option anymore.
  • Dedicated MIDI control bus switching fixed. Was closing the wrong bus eventually and crashing the whole show with it (fixes bug #2989590).
  • MIDI bank/program backout has been corrected on MIDI track properties dialog rejection (ie. user cancellation).
  • MIDI bank select method has been corrected for tracks with no instrument defined (probably fixing bug #2987071).
  • LV2 UI Instance and Data Access extension support added; reduce LV2 external UI parameter value update flickering.
  • JACK session infrastructure support. (EXPERIMENTAL)
  • Initial widget geometry and visibility persistence logic has been slightly revised as much to avoid crash failures
    due to wrong main widget hidden state.
  • Initial mixer widget extents are now set reasonably larger.
  • General source tree layout and build configuration change.
  • Ever since smooth-ramping introduction that having at least one input-only buses were causing immediate playback crashes, now hopefully fixed.
  • Refactored for common engine client nomenclature, primarily provided by JACK, then secondarily passed to ALSA Sequencer, getting rid of the JackUseExactName requirement and lifting the unique/single instance restriction in the process.
  • Current JACK Transport, MMC Device, and MIDI Song Position pointer (SPP) control modes are now saved/loaded as part of session option properties.
  • MIDI clip editor's context menu crash on Qt >= 4.6 has been fixed (resolving bug #2972603).
  • An ancient double-free corruption has been finally fixed at the audio/MIDI bus connection persistence logic.
  • Improved visibility of track state buttons text (R, M, S) when turned on dark colored themes.
  • LV2 Save/Restore extension support kicks off.
  • MIDI engine read-ahead period has been shortened to half than it was since inception--now it's a 500msec cycle.
  • MIDI clip editor event list gets its due inline editing, for time, note, value/velocity and duration columns, just
    one double-click away over the target cell ;)
  • Add-plugin selection dialog position and extent are now remembered across invocations and application sessions
    (tipping by Frank Neumann).
  • MIDI clip time-stretching is now made available through the same gestures as audio ones, by just shift+dragging
    either of the clip edges.
  • Drag-and-copying plug-in instances (cloning) is now fixed with regard to parameter value replication.
  • MIDI clip editor snap-per-beat setting is now independent from main multi-track view; File/Save As... dialog fixed;
    the current event selection is now kept floating as long as it's possible after editing command actions; finally,
    edit mode has been extended to free-hand event drawing, chalking off (piano roll) draw mode from the TODO list.
  • Swing-quantize has finally made its overdue debut as an additional MIDI clip editor tool (see Tools/Quantize...).
  • Almost since its inception, audio inserts were injecting garbage random noise when not being activated, now fixed.
  • Dedicated audio output ports for MIDI track plugins, now have their connection persistence back in business due on session load.

Cheers && Enjoy (what else?)


FYI, packages for Karmic are available now too:



rncbc's picture


Error it's always a mixed feeling.

We (may I?) *nix old-farts usually dismiss most problems as PEBCAK but, quite really, problem is in the "we" part of the process. Yeah. most of the times "we" assume that B is doing it, than C will be fine. Eventually, when D gets to it plain wrong, "we" always forget that it is A who ran the whole damn thing havoc. :]



Just backing up a moment, Rui, where you said in your release notes:- "Almost since its inception, audio inserts were injecting garbage random noise when not being activated, now fixed."

There's a classic example in itself! To you, that's a bug - to me. it was a feature, being the only way for a while I could tell whether a plugin was even communicating or not ;-) I even used it creatively on one track ;-)

Perhaps the mos'excellently cool and helpful AutoStatic/Jeremy could let you know just how he's got 0.4.6-4 running without problems - there, maybe, lies the key to why us lesser mortals have hit a brick wall?

QED. Your .debs _do_ work for someone or did he compile from source? {;-)

Hello danny_g, I use my own packages from my own PPA. So those packages are compiled from source. I don't use Rui's packages since those are for Lucid and I don't use Lucid (yet).



As promised, here are the ubuntu 10.04 packages, with everything enabled, except for the (experimental) jack-session:

@Rui: Please update/replace the ones on Qtractor website, as they are totally useless at the moment...

Thanks for your great work!

rncbc's picture

@falkTX: thanks. links are being promptly updated:


Co-operation in action. Love it! ;-)

Thanks so much to both of you. Roll on Qtractor in UbuntuStudio's Synaptic Main!

has anybody ever compiled this successfully on a mac? i get the "Qt library version >= 4.1... no; Qt 4.1 or greater is required" error no matter what i do. i have qt 4.6 installed. tried macports (also the devel version) and the dmg. tried to all kinds of --with-qt folders. is there a definitive guide on installing on a mac? i'd like to compile from source though, in case a package is planned.

rncbc's picture

May I assume you're asking for a MacOSX build ? If yes, then you better know right now that's NOT going to happen, no matter which Qt version you're into.

Qtractor is currently a Linux only application due to its hard-dependency on ALSA for MIDI infrastructure.

Sad but true :)

Thanks anyway, for asking

qtractor is quite impressive. It's a pity I cannot use it with files bigger than about 350 kB in size, in one single track - this is from straight midi recording, in time scales beyond an hour.

I waited, if this weakness would be taken care of, at some time during versions progress - but maybe I am the only one encountering this situation. In fact, this problem appears as a bug: In that case, import of the midi-file takes much longer, and after that, the file is kind of folded or wrapped, so different parts of the file (some two or three) appear like being merged upon each other. so the track is garbled. So, on first sight it looks like kind of memory overflow.

Interesting enough, having compiled qtractor freshly, this problem is not there immediately, but it appears after its third or so use and remains after that. I tried chasing the directly associated configuration and history footprint, but could not see on that level, where the problem is introduced.

This situation is stable behavior since quite some kernel versions now, and slackware-current has distributed quite some upgraded library versions meanwhile.

qtractor would be quite a bit more modest in use of resources than rosegarden, which works as expected.


Add new comment