New test version

All about the development and use of SynthFont2
Post Reply
chasp
Posts: 10
Joined: Tue Mar 17, 2015 8:39 pm

Re: New test version

Post by chasp »

Yup .. seems i have a firm grasp on the obvious but thought i should mention it....!
On the Piano Roll and the virtual keys ... first time I used them the first key I clicked on stuck and the program hung..... but strangely upon restarting Synthfont 2 clicking on a note cause no sound ... but running down the keys with right mouse button pressed caused multiple notes to sound and keep sounding until I clicked on a single key which made the sounding notes stop but SYNTHFONT 2 did NOT hang... hmmmmm. This must be frustrating for you Kenneth.....
chasp
Posts: 10
Joined: Tue Mar 17, 2015 8:39 pm

Re: New test version

Post by chasp »

Hmmmm not sure how to even explain this one.....Tracks 02 through 17L contain a mixture of soundfonts and VSTI instruments..... the arrangement i am using for testing seems to sound ok but Tracks 12 thru 17L SHOW no activity as to number of notes being played... the tracks produce sound but the NUM NOTES values stay at zero and there is no activity showing for the color activity bar in the name field for those tracks
More weird stuff in that now ( at least this time) the VIRTUAL KEYBoard is working for the tracks where no activity is showing and also for all other tracks in the arrangement...
hmmmm I dunno.....Obviously something strange here....!
Sorry Kenneth to be the Grim Reaper with this strange stuff....
Please let me know what i could do .. if anything... to help you on these things....
Chasp
Admin
Site Admin
Posts: 671
Joined: Tue Mar 17, 2015 9:22 am

Re: New test version

Post by Admin »

Chasp, maybe you could send me the arrangement (and the midi file) to make it easier for me.
Admin
Site Admin
Posts: 671
Joined: Tue Mar 17, 2015 9:22 am

Re: New test version

Post by Admin »

Here is the next one:
https://dl.dropboxusercontent.com/u/173 ... t2012Z.rar
  • Cuing was indeed pretty broken. I mentioned in a previous post that the trick is to get three threads synchronized, but in fact it was one more - the main thread (and there are more for multiple CPU cores, but this was not part of the problem). I think I have managed to synchronize all four now.
  • Virtual keyboard issue 1 - In ray890's report there was one single note still in KEY DOWN phase when mouse button was released. I had indeed added a call to a function to kill all notes, independently on phase, and spell out some data for the notes. Thanks to the debug output I may have found a possible culprit. This version does not KILL all playing notes, it only puts all notes still in KEY DOWN into the KEY UP phase (and tells about them in the debug output) , so they should keep playing for the duration of the release phase. So, please run again with DBWIN32 and if you experience notes still in KEY DOWN phase then send the output to me.
  • Virtual keyboard issue 2 - while the keyboard would instantiate notes for all layer tracks as well, the NOTE OFF event didn't do so. Hence, at least layer track notes would keep playing indefinitely.
  • Chasp had two issues with the MIDI Events list: 1) when the list background was supposed to be dark, it wasn't so. It remained white. Text color was also white, so nothing was visible. 2) switching theme would empty the dropdown list for selecting tracks.
  • Chasp had also an issue with activity not displayed in the tracks list while playing. As I almost thought initially this had to do with using layers. Thanks Chasp for sending me the arrangement so I could verify and fix the flaw.
  • That's all for now!
ray890
Posts: 40
Joined: Tue Mar 17, 2015 4:27 pm
Contact:

Re: New test version

Post by ray890 »

In this version, song cuing under regular usage appears to be in working state upon first look, and cuing is fast under External MIDI as well. The MIDI-Out still has the same PPQ bottleneck issue, perhaps that could be looked into with the next version.

However with cuing, stability issues arise (crashes, hanging, and/or error) once I start cuing forward to by repeatedly clicking the controls, especially in multiple channel MIDIs. If that doesn't reproduce the issue, then rapidly clicking and switching back and forth "forward to next bar", "to marker", and backwards will eventually trigger this. Also this seems to happen most frequently when gm.dls is used as the sound file. Another side-effect I found is that sometimes doing this will produce a sustained note that suddenly gets killed 3 seconds later.

For the virtual keys, there is no more infinately held notes, however I've noticed sometimes while dragging my mouse around it I've gotten an error report message and sometimes minor stability issues. Here's a sample output of while this has occured: http://pastebin.com/1ZZt2fHx
Admin
Site Admin
Posts: 671
Joined: Tue Mar 17, 2015 9:22 am

Re: New test version

Post by Admin »

Thanks!
Here is the first Release Candidate of 2.0.2.0:
https://dl.dropboxusercontent.com/u/173 ... t2020A.rar

I really hope this will be the last test version as well!
  • I've made a few changes to the cuing function again.
  • The source of the exception experienced by ray890 while playing on the virtual keyboard has been removed.
I will have very little time to work on software during the next three months so I will have to finalize this version very soon - before beginning of June.
ray890
Posts: 40
Joined: Tue Mar 17, 2015 4:27 pm
Contact:

Re: New test version

Post by ray890 »

I've gotten some extra tricks up my sleeve to help me find more stability issues and I've found 6 more bugs, after the most important of these get fixed I think we are ready to make it the official 2.0.2.0 release!:
Also, here's a re-cap of everything else I have in the "Not yet fixed" list for reference: Edit: Oh and, I would also like to see a toggle option under "view" that will let me disable the "mixer" panel at the bottom right as I never use it personally.

Edit #2: Added more findings and demonstration links to each bug listing.
Last edited by ray890 on Fri May 29, 2015 2:32 am, edited 16 times in total.
Admin
Site Admin
Posts: 671
Joined: Tue Mar 17, 2015 9:22 am

Re: New test version

Post by Admin »

Thanks ray890,
I will be out of office most of this week so I will appreciate it if you all can report as much as possible of your findings during this week. Next week I'll try to finalize version 2.0.2.0, but [unfortunately] I will be out of office again the later half of the week, so I may run out of time.
Admin
Site Admin
Posts: 671
Joined: Tue Mar 17, 2015 9:22 am

Re: New test version

Post by Admin »

Here is 2020B:
https://dl.dropboxusercontent.com/u/173 ... t2020B.rar
  • Deleting a massive amount of MIDI events can cause the program to crash due to running out of undo memory. -- I haven't been able to fully reproduce this problem. However, there WAS a flaw in the code that would make SynthFont believe it had got a new undo buffer when out of memory happened. The out of memory exception was caught by the code (message "Cannot create undo data - out of memory" displayed) but due to the flaw another real error appeared: "List index out of bounds". Fixed now, I think.
  • MIDI Notes are undeletable from the event editor if you click on the NoteOff events. -- Additionally, you could not edit the note either (or make a copy). Fixed.
  • Pressing Virtual Keys with keyboard keys trigger an error message and infinitely sustained notes. -- Fixed.
  • Hanging occurs if you happen to click the "stop" button before the "sorting events" stage. -- Not Yet Fixed. I will probably have to leave until later as it seems difficult to reproduce.
  • Clicking "Forward to next bar" rapidly during Black MIDI playback can cause hanging, crashing, or trigger an error message. -- Well ... I don't know about this one, but, there was a bug in the code used to go backwards.
  • Cuing backwards from a lag point can cause infinitely sustained notes until cued again, or an error. -- See above.
  • Changing song tempo during playback causes playback progress indication to become inaccurate. -- Well ... I don't know. Seems hard to reproduce, or then i am looking at the wrong indicator.
  • Reaction to Pitch Bend events are a little too strong. -- I guess you are actually talking about the Portamento effect? In that case I disagree, at least for now. I will check the code again before release, but I think you experience a difference between how SynthFont and other players implement the Portamento speed, for which there is no standard.
  • MIDI input monitoring mode - Integrate SynFonOne to replace current functionality. -- Leave until later.
  • Percussion channels are treated as non-percussion when using a soundfont with no percussion presets. -- Fixed. Any missing preset will be fetched from the standard SoundFont delivered with SynthFont (GMGSx.sf2) or, if missing, from Windows' own GM.DLS.
  • Has a track limit of 255. -- Leave until later.
  • Help function is unavailable. -- Leave until later.
  • Shift-clicking the open program window from the taskbar does not open a new instance of the program. -- There seems to be a bug in Delphi's runtime library. I cannot do anything about this now. Leave until later.
  • Switching between large soundfonts during speaker playback causes wait cursor to remain until playback is stopped. -- Haven't yet myself been able to catch this one. Leave for now.
  • Switching between voice limit settings during speaker playback can cause program to hang. -- Fixed.
  • External MIDI Output lags very easily under Black MIDI workloads. -- Leave until later.
  • I would also like to see a toggle option under "view" that will let me disable the "mixer" panel at the bottom right as I never use it personally. -- There is a "Netscape" splitter that you can use for this purpose. There used to be a standard splitter there, but you could not hide the mixer with it.
ray890
Posts: 40
Joined: Tue Mar 17, 2015 4:27 pm
Contact:

Re: New test version

Post by ray890 »

Oh, I should have mentioned that I've made a significant revision to my above post earlier today, with fixes of any misinformation on my findings and new findings, as well as links of videos/screenshots of each bug being reproduced; I encourage you to view the videos on all of the bugs your not familiar with. I've revised the post again to show which bugs have been fixed by 2012B.

2012B didn't fix or do anything to the crash caused by making a lot of undo points, nor the piano-percussion bug actually.

In this version, I see that no "Note Off" events are visible if Note On" is not checked, my idea is for "Note Off" to become a disabled checkbox if "Note On" is unchecked, unless you have a better way to deal with it.

About the "hanging if you click stop before sorting event phase comes" bug, I don't expect the typical person to humanly run into this issue, but I mentioned it anyway because it's still a stability bug introduced in a recent beta build nonetheless.
To reproduce it, I used "Free Mouse Auto Clicker.exe" to automatically click the play/stop button. To use it, change "seconds" to 0, and "1/100 S" to 1, and don't click start. Once the mouse is positioned on the button in synthfont, press CTRL+F2, and be absolutely sure you press CTRL+F3 to stop the autoclicking before you move your mouse off of the button.
Post Reply