Did I mention this one yet? Sometimes if I open up the Instrument editor while a song is playing, one of the song channels goes completely mute. Muting/unmuting the channel doesn't fix it (only stopping the song / hitting F12 does).
I still don't have a pattern for exactly when it occurs ... it seems more likely to occur on complicated tracks with a note every 2-3 rows but I can't pin down the specifics. It also seems to occur only on specific instruments like 06 and 07, not 00 or 01, and Triangle is especially susceptible to it....
[EDIT] I MAY HAVE FIGURED IT OUT!
- Open the attached file and start playing the song (note this is just an excerpt to isolate the problem; it's not the full thing).
- With the song playing, move the cursor to Pulse 1 and double-click any instrument to open the Instrument Editor.
- Click anywhere inside the Instrument Editor that is empty space -- e.g. to the right of the DPCM tab, or between the tab control and keyboard -- and the current channel goes completely silent.
- This lasts until the next volume change (which in the case of Pulse 1, occurs at rows 00 and 60 of the pattern).
- Now move the cursor to Triangle and repeat the above steps (open Instrument Editor while playing, click empty space); Triangle goes completely silent now. In this case the channel stays silent no matter how many times the pattern loops (because there are no volume changes in its pattern, as Triangle is a fixed-volume channel)
- Hitting any note on the Instrument Editor's keyboard restores the channel sound to normal (after the note has finished playing, of course). Clicking other controls in the Instrument Editor has no effect. Instrument changes (such as row 00 of Pulse 1, or anywhere in the Noise channel) have no effect. Muting/unmuting the channel has no effect. Killing the sound (F12) or restarting the song (F5/F7) works.
If I have a song that starts with a pickup, I'll usually ID the pickup measures as (starting from) 7F so that pattern ID 00 is the starting point of the song's loop.
This creates the occasional issue where for some reason FamiTracker will never actually assign pattern IDs of zero to a new frame, even when they're unused.
E.g:
- Start a new file.
- On the frame editor, manually relabel all channels as pattern ID 7F.
- Insert a new frame; it is assigned pattern IDs of 01 (even though 00 is available).
za909: I think I found the cause of that crash, it should be fixed now.
Stratelier: I found the cause of the channel mute problem, thanks for the instructions for how to reproduce it. Clicking anywhere in the instrument area will send a note cut command, but also sets the volume to zero. That was the problem.
I've attached a beta build of the next version here, I'd just like know if any of the win XP users still have problems with missing icons? I assume it's fixed now but would just want to double check.
Is the choice of absolute/relative/fixed arpeggio instrument sequences made per-module or per-instrument? I've run into a minor snag but can't call it a bug unless I know what the standard is.
Okay, thanks. On closer inspection there's no actual bug here, I just wasn't paying close enough attention the first time around. I actually like how a fixed arp plays the sequence then goes back to the intended note....
Okay, I found a new one:
1 - Start a new song with four frames (patterns 00 through 03, in that order).
2 - On the Frame editor, jump to frame 03, then right-click frame 03 and tell it to Insert Frame (you get frame 04 with pattern IDs = 04; as expected).
3 - Right-click on frame 05 and tell it to Duplicate Frame (you get frame 05 with pattern IDs = 04; also as expected).
Now for the buggy behavior.
4 - Right-click on frame 05 (pattern IDs = 04) and tell it to Remove Frame. Frame 05 is removed and the selection cursor is automatically moved to frame 04.
5 - Now right-click on frame 04 and select Duplicate Frame. This creates frame 05 with pattern IDs = 00, not 04. It's as if FT tried to duplicate frame 00 instead of 04....
6 - Remove frame 05 again (selection cursor moves back to frame 04 automatically).
7 - Right-click on frame 04 and tell FamiTracker to Insert Frame. You also get a frame 05 with pattern IDs of 00, despite that pattern ID 00 is already in use (in frame 00; whether or not pattern 00 is empty is irrelevant)
And for some further illumination:
8 - Remove frame 05 (cursor jumps back to frame 04 automatically).
9 - Jump to a different frame (e.g. click on frame 00) and back again to frame 04.
10 - Now right-click and tell FamiTracker to Insert (or Duplicate) Frame; you get frame 05 with pattern IDs = 05 (or 04) as expected.
EDIT: I just checked out the beta release above, and the below bug is fixed. I must have missed an above mention of this bug because I didn't exactly understand what was being said. :P
Found a bug with the release note function (===). 0.4.3 will repeatedly play the release envelope multiple times after the first, rather than playing it once and ignoring the rest. For example;
{00 : E-4 00 . ...}
{01 : ... .. . ...}
{02 : ... .. . ...}
{03 : ... .. . ...}
{04 : === .. . ...}
{05 : ... .. . ...}
{06 : ... .. . ...}
{07 : ... .. . ...}
{08 : === .. . ...}
{09 : ... .. . ...}
{0A : ... .. . ...}
{0B : ... .. . ...}
Above, in 0.4.3, will play the note at row 00 with any attack, loops, and what have you, then will play the release at row 04. However, at row 08, it will again play the release section by itself.
Attached is an example .ftm, and two .wav outputs from 0.4.3 and 0.4.2, in case it's somehow my computer. The bug is not present in .2.
- Open up any new/convenient song and start editing.
- Select a channel and press the Q key (high C) and while the key is held down, hit Tab; then release both keys.
- The note is still playing.
More technically: When you press a key, Famitracker starts playing the note, and when you release that key, Famitracker cuts the note, right? Well, if you happen to hit TAB during this time (e.g. maybe you fat fingered a high C) which causes FamiTracker to from channel A to B, it ends up cutting channel B instead of channel A where the note was actually played. If the bug is reproduced multiple times across different channels you can tell FamiTracker is cutting the wrong channel.
Techdisk wrote:
Found a bug with the release note function (===). 0.4.3 will repeatedly play the release envelope multiple times after the first, rather than playing it once and ignoring the rest.
In 0.4.2 at least, if you have a multiple release commands any part of the release sequence that hasn't actually finished yet will retrigger but other parts will not.
I've always found the TAB thing useful myself. It's pretty neat being able to preview multiple notes at once. And you can always hit F12 to kill every note.
I notice that if you resize FT's window, the menubar wraps but the toolbar does not, and that is a problem. (Tested - FT 0.4.3; Windows XP +SP3)
On a tangent, there seem to be no enforced minimum window dimensions whatsoever (I can resize the window down to 112x27 if I so choose, at which point nothing is visible except the window border provided by the OS). Now in a practical sense, the minimum window dimensions (to display all 5 core channels with 1 fx column simultaneously, and display 16 rows in the pattern editor) are about 750x546 (Frame Editor Left) and 595x570 (Frame Editor Top). So maybe FT should enforce minimum window dimensions of around ~750x570?
Oh, speaking of channel fx columns, sometimes I would like the ability to hide a channel from the display entirely. Or at least shrink the channel width (subsequently suppressing the fx/volume/instrument channels in that order).
Exporting to NSF, I noticed something that probably never meant a problem to people, because it's not a very common feature to use.
It seems the DPCM loop flag is never set in .nsf files. This also happens with 0.4.2 and 0.4.4 beta exports.
