on the TB-303, when the patterns are chained, and you edit them that way - they are treated as one single pattern, unused notes are not taken into account
thus, if you remove 1 Gate from the first pattern - the note that was associated with it is now shifted <forward> to the next step with Gate, thus all following notes too, and as a result the very last note from this whole buffer becomes unused (and will be lost when this thing is saved)

in other words, say you have 2 patterns selected
they both contain 16 Gates, the first one is full of c# notes, second one is full of d# notes
now if you chain them and replace the Gate on step 5 with a Rest - Pattern2 will begin with a c# note followed by 15 d# notes
see, it effectively shifts ;]

wow. i can see it presents a problem!
i was wondering how you obtain your 'shift' <L R> of
the sequence in beta2 (and 1). there must be a stage
between sequencer and the note slot it is addressing.
(it's a bit like an arpeggiator, isn't it? add to list,
remove from list...does DH stack/queue etc do things like this?)
but i wonder what method you use for this. is it something
like a text array?
in any case, what you are trying to do seems pretty tricky.
once again, you -ahem - could always release an interim beta3
while you work it out!

the problem is when i try to preserve those unused notes, the TB-303 simply ignores them

the pattern manipulator Shift < > is pretty simple

DIR = -1 for shift forward > and 1 for shift backward <
while (i < len)
{
    i2 = (i+DIR+len)%len;
    tmp[i] = buf[i2];
    ++i;
}
// copy "tmp" into "buf"

i have made a much more complex variant of it in the current sequencer, which can only shift given attributes (via a mask) say.. shift only the Key+Acc+Up
this mask thing applies to nearly all other manipulators (reorder, randomize, mutate..)

no idea about the DH modules you mentioned..
these things are better done in C++ directly
...the whole sequencer actually, i can't imagine how you can do this with standard/3rd-party SE modules, maybe you can, but they will have to be very specialized which means you will probably need to get coding yourself, which i went thru as you can see

ahem indeed. Go on Antto....

lol, yeah, fair enough

yo bump or you're gonna drop off the front page.

still haven't fixed the issue in the sequencer (still no 303 available for the test i need) so i am waiting on that one
meanwhile, i've been working on the synth part.. solved the cutoff envelopes entirely (including accents and what not)
i took the x0xb0x waveform and use it to sync Venom's oscillator while the x0x also controls Venom via midi
this way both synths "play" the same thing BUT the waveforms of both also is in sync on the scope, which makes it MUCH easy to do the things i have to do normally
so i managed to aproximate the audio path (filter->vca->output) much better than my previous aproximation, and recorded this little video to show the funky setup:
http://www.youtube.com/watch?v=pnBEACajFtg
it's a bit chaotic but basically look on the scope and on the sonogram

afterwards, i gave another try on the square wave, this time using this oscsync setup, tried many things and finaly came up to this model: https://www.box.com/s/dmtiwc5kjy5y6jv9e9oc
this recording is split in two parts (there's a beep in the middle) i will not tell you which is which..
if anyone wants to listen more seriously to it - download it, because the online player plays a converted (to mp3) version of the wav..

afterwards, i noticed something small, and changed the filter/audiopath and envelopes yet again, it's a well known thing in the x0xb0x - a small dc offset before the vca, i managed to aproximate the signal path even better now

this all means that Venom will be more like an aproximation of my x0xb0x (soundwise) .. you decide if that makes it a 303 or not enough
i've been very happy since i got this x0xb0x so it's good enough for me to have another one, with a "better" sequencer (which is currently broken, huh)

so it looks like it won't happen this year, at least i will *hopefully* have my tests performed till the end of december, so i will be able to figure the issue and move on with the sequencer...

got my tests performed
my initial idea about "preserving the unused notes" collided with the normal operation of the TB-303 when multiple patterns are edited at once (aka chained)
i also had the wrong impression that the TB-303 pitch buffer is 64 bytes while it's actually only 48 (this means you can't really edit a chain which has more than 48 notes in it)

so i let my idea go (about preserving the unused notes) and "fixed" the sequencer, now it behaves like the TB-303 (apart from the pitchbuffer which i keep 64 bytes long, thus you can safely edit 4 patterns)

this means that some/most/all of my pattern manipulators will no longer work correctly, i will have to pay them a visit later on..

i also changed the pattern selection, it was like on the x0xb0x - 16 banks of 8 patterns.. but there is no longer a need for this
so i did it like on the 303 - 4 banks of 2 sections (A, B) of 8 patterns
it's still 128 patterns, but there's a twist with the section thing
the A/B can be used as a "pattern variation" .. for example, you can program 2 patterns in section A and make a variation of them in section B (on the same pattern slots)
then you can chain them and transpose them in Play mode, and then, hit section B - you get the "variation" of those two patterns without this breaking your chain, nor reseting the transpose.. i noticed there is something similar to this in the 808 sequencer iirc

and now i'm looking forward into implementing track mode (like on the 303)

this is tap-write mode with chained patterns:
http://www.youtube.com/watch?v=9DE28qeldII
notice that pattern3 is triplet, this is no problem of course ;]

Probably already mentioned in the 50 pages previous, but will there be a Mac version of this?

As this is made using SynthEdit, no.

Hi,
I translated the VB-303 manuals into Japanese, because no one else did it.
If you have any problems, please let me know.
http://blog.livedoor.jp/acid808/archives/21660372.html
Thank you.

yazzz: cool

thisplace: nope, no Mac version

haha, yes, and antto is going to do a website this time

happy new year

what website? ;P~

i made a drastic change in the sequencer memory again
previously, i used 999 DUAL INT pins which meant 3996 bytes of memory, but there was quite a long initialization time (that was also since beta1)
it wasn't fair because the x0xb0x has 4096 bytes external + 512 bytes internal eeprom
but instead of adding more INT pins (which was the easiest thing) i changed it to DUAL TEXT pins, 72 of them, and i uses a very simple encoding to prevent localization issues (since this is SDK2, SDK3 has dt_blob)
and so each string holds 128 chars, which hold 64 bytes of raw data, but less memory is wasted around the actual pin properties structures..

as a result, it now has 4608 bytes of total memory, and loads uber fast
it will be enough to store all 128 patterns + 16 tracks + additional settings (per instance)

the distortion unit would have to be on the "rear" panel, and i haven't figured out how it would look like yet, it would have about 5 knobs..
i'm thinking about something green-ish with black knobs

lol, the idea of you sitting there creating 999 pinouts...

the distortion unit could be like one of those Boss boxes.
different colours, different flavours of distortion ???
(currently loving Pearl Thriller multipeak with pre/post
distortion. it is a 6x res peak BPF, or something like that)