ChatGPT can do Lua #553
Replies: 69 comments 107 replies
-
|
oh dear: something in my original script is causing weird text issues. hopefully you get the general idea. (and when i copied it into the panel, the indents were all good, not as above) and is its comment/explanation: "This uses a table tabValues to map the values of structure to the corresponding uiTabsCurrentTab values for SynPcm1 and SynPcm2. This allows us to eliminate the multiple if statements and reduce the code to just a few lines." |
Beta Was this translation helpful? Give feedback.
-
|
Hi subvOx - that code is unreadable - try formatting it with https://goonlinetools.com/lua-beautifier/ and place the code between <pre></pre>tags! struct12_showGrps_pt1 = function(mod, value)
local structure = panel:getModulatorByName("struct12_pt1"):getValue()
local SynPcm1 = panel:getComponent("tabs_SynPcm1_pt1")
local SynPcm2 = panel:getComponent("tabs_SynPcm2_pt1")
local tabValues = {
[0] = {0, 0},
[1] = {0, 1},
[2] = {1, 0},
[3] = {1, 0},
[4] = {0, 1},
[5] = {1, 1},
[6] = {1, 1},
[7] = {0, 0},
[8] = {1, 1},
[9] = {0, 0},
[10] = {1, 0},
[11] = {0, 1},
[12] = {1, 1}
}
SynPcm1:setPropertyInt("uiTabsCurrentTab", tabValues[structure][1])
SynPcm2:setPropertyInt("uiTabsCurrentTab", tabValues[structure][2])
end
|
Beta Was this translation helpful? Give feedback.
-
|
lol sorry it is isn't it. i'll have another go. |
Beta Was this translation helpful? Give feedback.
-
|
woo got there in the end. my very bad habit of decorating methods with separators. bear in mind this was my first attempt at doing anything in Lua long before our discussions! - i think chatGPT did pretty well. |
Beta Was this translation helpful? Give feedback.
-
|
in fact, it looks like it's made a mistake: the first 2 entries are supposed to call the first sub function see(00) eg: result is (0,0) (0,0), not (0,0) (0,1) what would dnaldoog's version be? |
Beta Was this translation helpful? Give feedback.
-
|
I'd probably do it the chatGPT way using table matrices, but your way would work too l imagine :) |
Beta Was this translation helpful? Give feedback.
-
|
it does correctly identify that using a table is better. |
Beta Was this translation helpful? Give feedback.
-
|
A slightly more concise version of your code would be to have only one internal function and set it to local: -- Called when a modulator value changes
struct12_showGrps_pt1 = function (mod, value)
-- sets corresponding option views
local structure = panel:getModulatorByName("struct12_pt1"):getValue()
local SynPcm1 = panel:getComponent("tabs_SynPcm1_pt1")
local SynPcm2 = panel:getComponent("tabs_SynPcm2_pt1")
local function see(a,b)
SynPcm1:setPropertyInt("uiTabsCurrentTab", a)
SynPcm2:setPropertyInt("uiTabsCurrentTab", b)
end
-- 00_01_SS
if
structure == 0 then see(0,0)
-- 01_02_SS
elseif
structure == 1 then see(0,0)
-- 02_03_PS
elseif
structure == 2 then see(1,0)
-- 03_04_PS
elseif
structure == 3 then see(1,0)
-- 04_05_SP
elseif
structure == 4 then see(0,1)
-- 05_06_PP
elseif
structure == 5 then see(1,1)
-- 06_07_PP
elseif
structure == 6 then see(1,1)
-- 07_08_SS
elseif
structure == 7 then see(0,0)
-- 08_09_PP
elseif
structure == 8 then see(1,1)
-- 09_10_SS
elseif
structure == 9 then see(0,0)
-- 10_11_PS
elseif
structure == 10 then see(1,0)
-- 11_12_SP
elseif
structure == 11 then see(1,1)
-- 12_13_PP
elseif
structure == 12 then see(1,1)
end
end
|
Beta Was this translation helpful? Give feedback.
-
|
yes in fact chatGPT mucked it up because was only using [1] & [2] |
Beta Was this translation helpful? Give feedback.
-
|
oh i understand what that is now: it's the table: is each entry a 'subtable'? so it has index 1 and 2. that must be it. |
Beta Was this translation helpful? Give feedback.
-
|
and in the original table, it had fetched all the options i had listed - it got them all correct except the second one. |
Beta Was this translation helpful? Give feedback.
-
|
ok just had another go with something else. got the Lua Beautifier standing by. `-- modulators -- tvaLevel zero is same as partial Off, AND with partial onoff -- components -- ledTextB 2nd row//shows partial is at both inputs// -- tab OnClick buttons//showhide with partial onoff -- tab showhide -- struct12/34/clikStrct12/34//showhide for partial pairs |
Beta Was this translation helpful? Give feedback.
-
|
..not having a lot of luck with the formatting, but it's broadly there. so here's what chatGPT did to that: |
Beta Was this translation helpful? Give feedback.
-
|
it messed up the last 2 because the naming 12/34 was confusing. of course it doesn't know the difference between getModulatorValue() and getValue() ! (and i don't remember, myself) here's the rest, picking up from last lines above; |
Beta Was this translation helpful? Give feedback.
-
|
yeh well it's wrong, but it attempts an approach using tables and index again. (i said it was wrong yesterday, and realised later it wasn't..) - can now see some more mistakes. but it has suggested an alternative approach, which is interesting, and can sometimes 'see' through, enough to do some optimisation. |
Beta Was this translation helpful? Give feedback.
-
|
finally, for now, here's another bit, defining colours for a LED pair (i'll explain what it is afterwards). there's quite a nice form here in the first bit: |
Beta Was this translation helpful? Give feedback.
-
|
anyway: this is from my old D-110 panel, that i was looking at again yesterday. i was able to optimise a pair of methods very nicely - which made me go through the whole thing again. (i haven't used Ctrlr for ages, since early last year.) (and i was just thinking what fine work i'd done, sorting out those button and LED operations ! :D ) |
Beta Was this translation helpful? Give feedback.
-
|
i think chatGPT is a scam haha. it's some kind of callcentre with loads of human chatbots, scratching their heads when you ask them to code something. no actually, it was absolutely awesome just now, when i asked it to swap a protracted longform elseif series into a more concise form related to the 4 bit binary count represented by 4 button states, 4 parameters grouped on one parameter number - anyway, it came back with the same reply dasFaker gave me years ago lol. |
Beta Was this translation helpful? Give feedback.
-
|
it's a bit lazy, can't be bothered to give a full list of midi messages for 00 to 0F, ... just gave me some dots, and 'finish it yourself'. |
Beta Was this translation helpful? Give feedback.
-
|
here's more context with an earlier workfile. still has old code in. this is building out the structure of the panel, making methods apply to all parts, and doing part1 before fleshing it out. this is where i should start the new thread :-) |
Beta Was this translation helpful? Give feedback.
-
|
ok started a thread for the D110. have much waffling to do there. meanwhile, new conundrum is how to resolve the 'partial reserve' in the Patch parameters. with a total of 32 partials, it is possible to assign a voice reserve for each part. but it can't allocate more than is available depending on how much is already allocated to other voices. there are 8 parts, and a rhythm part. partial reserve can be from 0-32. (if you set 32 on one part, no reserve is available for other parts) is that a job for math.floor or math.ceiling? (etc not sure what i'm talking about yet) |
Beta Was this translation helpful? Give feedback.
-
|
Can you post your work? It’s hard to understand what you’re trying to do. I think my latest TX81Z panel might have the logic you need; not sure ^_^ |
Beta Was this translation helpful? Give feedback.
-
|
with a quite minimal prompt, chat GPT gave me: i suppose it would look something like this: start file with 9x dropdowns 0-32 and a 'total' display. noted something interesting in the steinberg atari editor for D110, which was a XY mixer with the 4 partials, like k1/tg33. |
Beta Was this translation helpful? Give feedback.
-
|
just to follow up with this, i've done a bit more prompting with chatGPT and have a functional panel, that does NOT produce the correct results - but it seems to be going in the right direction. to recap, the device in question, D-110 has polyphony of 32 partials, and 9 parts, including the rhythm part, and you can assign a 'partial reserve', same sort of thing as voice reserve, for each part. notionally each part 'reserve' parameter has range of 0-32 but in practice, the control maxes out when the total partial count is assigned, so if each part already has 2, or 3, or 4 reserved, the control you're incrementing might only go up to 10, or less, depending on what remains of the total. i thought about controlling the modulator property "maximum value", and did get it working, but it doesn't update the list on a combo - don't know how one might update a combo list in realtime? nightmare; doable probably but maybe just use a better modulator type - and really, i'm trying to prompt chatGPT for a mathematical solution rather than resorting to Ctrlr parameters ( which won't help with the UI .. i thought dropdown would be fine; fader, knob, meh not really. code below generated by chatGPT, from description prompt only, with minor interference from me: |
Beta Was this translation helpful? Give feedback.
-
|
ok i have a semi-working voice reserve thing going on. and i've been looking at what happens on the D110 itself: it just doesn't offer any more incremements when max reserve is used up - which is hard to represent in Ctrlr; i've been using combo/dropdown list, which depends on the 'combo contents' component property. i'm not very keen on playing with Ctrlr properties 'dynamically' - i accidentally created properties that shouldn't exist yesterday, addressing modulator rather than component, and now can't get rid off them, without going on a detour to find out how to delete properties ... - BUT i have successfully altered modulator 'maximum value' on the fly - unfortunately, with the dropdown, that isn't enough because the list items don't change - so i had this mad plan to re-write the combo contents on the fly, mwah ha ha, and i think i can do that. it has to resemble what is in the Utility pane, namely each entry being separated by which in this case prints |
Beta Was this translation helpful? Give feedback.
-
|
ok i have it dynamically re-writing the combo list - lol what a mess. it isn't the Utility page format, it's just as it's written in the combo, with sorry!! have strayed from the previous discussion about the led display - because: i'm undecided whether to go with the one where the led just changes colour, or stick to the original dual led thing. plus i liked the code for both of those :-) i have the D110 next to me (with a very feeble backlight unfortunately on this one; gonna have to change it) (which is possibly just a standard OG 1602, which they only do in 'grey' or green these days, unless Circuitbenders.co.uk still do them modified with the new type, eg: more colours) - and the reserve display is just, like: 'part4 reserve: 3' and if there's no more partial RESERVE left, it won't go higher, only down. i thought the combo dropdown list would be the best way to represent it. or else i'll have to use a graphic strip slider. (much much much simpler :-) ) which means forget the setProperty thing ( as expected ) and try to figure out the maths: |
Beta Was this translation helpful? Give feedback.
-
You can delete a property you created with panel:getModulatorByName("lfo delay"):removeProperty("blah")
… or if you set it up as a part of the component() class: panel:getModulatorByName("lfo delay"):getComponent():removeProperty("blah")
… or maybe panel:getComponent("lfo delay"):removeProperty("blah")
… or in the callback (untested!): function x(mod,value,source)
mod:removeProperty("blah")
end
…or function x(mod,value,source)
mod:getComponent():removeProperty("blah")
end
|
Beta Was this translation helpful? Give feedback.
-
|
thanks, yes, i thought i'd seen something for that over on the JUCE pages. chatGPT has no notion of modulator vs component unless you explain it. i suddenly noticed oh i had the same property (uiLabelText) under both modulator and component. don't think it would be a big deal, only that no change happens if you address the wrong one. -positive experiment, however, since it can evidently be done. however it, and the updating needs to happen in specific order, so it needs a few tweaks. your version is pretty good, would need working out where the 'update' happens. but addressing properties like that doesn't seem very 'solid' - you could get stuck, having to reset defaults etc. - i'm wondering how Roland did it in their firmware. maybe something similar, with min/max default range. probably not very far off, with what chatGPT has been offering: add up current reserve values, deduct from default max for total remaining polyphony, .. not sure about divide by number of parts.. have to have another think about it. anyway, not urgent this bit, plenty to do in the meantime. this comes in at 'multi' level, the Patch. in fact i've been poking around the D110 interface - where i now pretty much can find everything quickly - and there's a page where it refers to 'Parts', where the reserve parameter is. (and so Timbre sits on a Part, within the Patch, and 'contains' the preset, called a Tone; i had some confusion over this, because you can access the Tone number and change it, without rewriting the Timbre first - which poses questions re: saving order. in fact i'm going to do a few experiments manually on the device to see what/how it happens.) (and then chuck the thing together and get started wrapping up the project). |
Beta Was this translation helpful? Give feedback.
-
|
You can write that function with less code :) dynaMaxTest = function(--[[ CtrlrComponent --]] comp --[[ MouseEvent --]], event)
local currMax = panel:getComponent("currMax"):getValue()
local newList = {}
for i = 0, currMax do
table.insert(newList, string.format("%d=%d", i, i))
end
local dataList = trim(table.concat(newList, "\n"))
console(String(dataList))
panel:getComponent("myCombo"):setProperty("uiComboContent", dataList, false)
panel:getComponent("myCombo"):setValue(0, false)
end
function trim(s)
return s:match "^%s*(.*%S)" or ""
end
|
Beta Was this translation helpful? Give feedback.
-
|
very nice ! |
Beta Was this translation helpful? Give feedback.
-
|
here's what chatGPT suggested, for avoiding addressing properties (such as max value): really minimal prompting required for this, and i was waffling a bit/being imprecise. ha, it also showed me the syntax for integer and replaced math.floor for me: and this is the next script i'm going to try out - i'm losing track a bit at this stage - i'm not convinced it should divide by 9 parts tbh; i think that's a false route i created with an earlier prompt. (i don't know how this works yet; i'll do the work a bit later) (yeh we don't know where it's going at this point ... ) |
Beta Was this translation helpful? Give feedback.
-
|
Here's a panel that might give you a head start on the max = 32 problem. I tried rewriting the function updatePartialReserve(mod, value, source)
if panel:getBootstrapState() then
return
end
local n = L(mod:getName())
n = tonumber(n:sub(-1))
counter[n] = value
local newsum = 0
for i = 1, 9 do
newsum = newsum + counter[i]
end
if newsum > 32 then
mod:setModulatorValue(0, false, false, false)
newsum = 32
else
local remaining = 32 - newsum
panel:getLCDLabelComponent("lcd_rem"):setText(String(remaining))
panel:getLCDLabelComponent("lcd_total"):setText(String(newsum))
end
end
counter={}
for i=1,9 do
counter[i]=0
end
|
Beta Was this translation helpful? Give feedback.
-
|
yeah that's nice. i was thinking of popup as alternative but i don't get have full control over colours, like background - text colour is ok, not sure if the font can be changed. i'm not especially attached to the display as is, it was just the quickest way to sketch the thing out. (so popup could do it) |
Beta Was this translation helpful? Give feedback.
-
|
had time to check closer and mess around with it. it's really good, and nice short method. that's a go-er, i reckon. sole quibble is how it resets to zero if out of range. but it will do the job in a panel for now. btw checked a default 'Patch' on the D-110 to see how it defaults the 'partial reserve' distribution (trying to get some insight into Roland thinking) and it's: 8,8,4,4,4,4,0,0,0 - that doesn't affect how this works; it's just what they saw as a suitable initial distribution, and will be the 'reset'. (the empty 'counter' method was abandoned when you put the counter in panelInit ?) the bit i'm looking at: so i shall attempt a collector of the array of current/last valid values, and try something like:
still looks like the combo might have to be substituted for another modulator type. (unless it ends up working as hoped for) |
Beta Was this translation helpful? Give feedback.
-
|
nope my plan didn't work. how would one return to existing value after going over the max value? if it's 4 and total max is already 32, and i go over, it resets to 0 rather than previous value of 4. (nb: wondering whether this is going to be confusing for user. meh as long as it works...) i got C errors due to |
Beta Was this translation helpful? Give feedback.
-
Ah - of course - resetting to 0 isn't good, but there's an easy fix - try this code: A new variable function updatePartialReserve(mod, value, source)
if panel:getBootstrapState() then
return
end
local n = L(mod:getName())
n = tonumber(n:sub(-1))
revertValue=counter[n] -- old value
counter[n] = value
local newsum = 0
for i = 1, 9 do
newsum = newsum + counter[i]
end
if newsum > 32 then
mod:setModulatorValue(revertValue, false, false, false) -- restore old value
newsum = 32
else
local remaining = 32 - newsum
panel:getLCDLabelComponent("lcd_rem"):setText(String(remaining))
panel:getLCDLabelComponent("lcd_total"):setText(String(newsum))
end
end
|
Beta Was this translation helpful? Give feedback.
-
|
aaah finally i understand 'counter' ! :-) |
Beta Was this translation helpful? Give feedback.
-
|
ok fiddled with that a bit, and have something odd going on displaying last values list - think it needs to initialise on each run but not sure how that should be done. thought i'd try to start working in your combo edit method, to see if that can be got to work (notwithstanding that an edited list comes up blank until a new selection is made - maybe that can be triggered? ) the small shortcoming of the previous panel is that if it goes over (total max), it goes to the last selected value - rather than zero, that's good - and not the last available value. i don't know how that could be achieved. i took out the indiv displays, and made the spare display show a 'reserveValues' table each time, which reveals the erroneous values. haven't actually achieved anything with this one, and the first attempt at messing around with substituting |
Beta Was this translation helpful? Give feedback.
-
|
just got a little gift from chatGPT to convert Roland preset count 11-88, to decimal 1-64, and back. first, decimal to Roland count (as chatGPT decided to call it) 👍 and then the other way: |
Beta Was this translation helpful? Give feedback.
-
|
oh ye-eah... just went back and asked it to extend it to cover A-11/88 to B-11/88 (total 1-128) and it came up with this: here's a little panel for that: don't know if it needs offsetting -1 to have 0-127 at this point. it's just a UI thing atm. |
Beta Was this translation helpful? Give feedback.
-
Hi @subv0x You can streamline that code a little with: rolandPreset = panel:getLCDLabel("rolandPreset") -- declare outside function
function converter(mod,decimal,source)
local prefix = "A-"
if decimal > 64 then
prefix = "B-"
decimal = decimal - 64 -- Adjust the decimal value for the second part
end
local row = math.ceil(decimal / 8)
local column = (decimal - 1) % 8 + 1
rolandPreset:setText(prefix .. row .. column)
end
|
Beta Was this translation helpful? Give feedback.
-
|
i think i had presented it with code originally in Java, by the author of EdiSyn, and the setText() bit with string format as 'sf' was yours :-D this numbering system is needed to select and display Timbres - would apply doing a Juno panel too ? - but it's going to cause complications with the actual handing of memory slots - which i've moved onto, retrieving bits we looked at back when i was saving sequencer patterns, if you recall (lol 2018 ... ). chatGPT suggested an alternative, not that dissimilar; i've botched something together that rewrites the table values, and retrieves them as text-to-label, but crashes sending the retrieved values onto modulators - using something that has worked before. have to check if it's because i'm using combos, and the lists are not numbered as eg: '1 = 1, 2 = 2 ..' etc. but simply '1,2,3..'etc. (laborious task that i've avoided but should perhaps try). i think the D-110 Timbre memories thing is a 'side-car', of those 7 parameters, which are also a subset of the Patch Part parameters, with a total of 11 parameters per part, stored in the 'Patch', which is the term used for 'multi'. the actual sound 'preset' is called a Tone (comprising Partials), which is assigned to a Patch Part, generally by calling up a 'Timbre' on that Part, or else by directly editing the Tone and ToneGroup (bank), and saving the Patch like that - both work for saving the multi setup. proviso being, that the 2nd way, it isn't saved as a 'Timbre' to be recalled and scrolled through. i think that's broadly how to understand it. thus, the 'timbre temporary area' must be the total space allocated for these parameters, as flash ram - because its retained through powercycling (as long as you don't refresh the 'Patch' memory) - and i suppose a timbre preset is loaded into it, in the same way you might make a manual edit to 'Part' parameters - it's confusing how you access these from the front panel buttons; some of the parameters are under 'Part' buttons, and some are accessible via Timbre>Edit - this suggests you'd want to save the Timbre as well, before saving the Patch; but you don't have to; you can just save the Patch itself (repeating what i wrote earlier to reinforce the point, because it's weird). at least this is how i think it works. 'Timbre' is like a library module/macro for a few parameters. this is where i've got to, writing tables but unable to restore saved values to the modulators - lots of stuff renn'd out that i've tried. look at 'popup' , saveTimbre, panelInit. haven't rewritten the combo lists yet. (the popup will also be retrieving 'tone' names, from the preset banks a,b,r (fixed tables), and from bank 'i'(internal user, also 'c' when card in use), which will depend on user tone memories table, which the popup will have to reference ... 8-0 |
Beta Was this translation helpful? Give feedback.
-
|
Does this panel do what you are wanting? Timbre Memories_5_0_test_dnaldoog_table_storing_values.zip I create a multidimensional table in --panelInit()
recorder={}
-- add any modulators to be stored here
-- when retrieving from D110 store values in table
for i=1,128 do
recorder[i]={
assignMode=0,
keyshift=0,
outAssign=0,
toneGrp=0,
bendRange=0,
toneNmbr=0,
fine=0,
}
end
-- store the value changed by user in each modulator storeValue = function(mod,value,source) local dec=decimal:getValue() local n=L(mod:getName()) recorder[dec][n]=value end -- recall the value in the table called from callback function in `decimal` `restoreValue(value)` -- which popup selection also fires function restoreValue(v) for k,v in pairs (recorder[v]) do panel:getComponent(k):setValue(v,false) end end |
Beta Was this translation helpful? Give feedback.
-
It all depends how you reference the modulator. In that code I just use you could now easily save that table as a JSON file and reimport anytime 🙂 |
Beta Was this translation helpful? Give feedback.
-
|
Here is the same panel but now you can save changes to table Timbre Memories_6_0_test_dnaldoog_table_storing_values_JSON.zip |
Beta Was this translation helpful? Give feedback.
-
|
ok i'm not sure what JSON is or what for ? is this for a librarian? could a panel write its own init/*.inf file for a preferred start-up eg: "do you want to save current panel config?" dialog on exit? also: how is also: the order of the keys in the
|
Beta Was this translation helpful? Give feedback.
-
it's called from function
Could be wrong but I think having a popup prompting to save data on exit() in Ctrlr is problematic or maybe not even possible; You would need a discrete button for that rather than have an automatic popup on exit. for example with VST instances it wouldn't really apply I am thinking, not sure there. Make it up to the user to decided whether to save data, unless you use the following: For save and load there is
|
Beta Was this translation helpful? Give feedback.
-
|
k for keys, ... and i never thought of 'i' as index, how about that. this is what happens when you fall down a rabbit hole thinking you'd just knock out a sysex control panel in Ctrlr. ok so no exit popup dialog, no problem. there could be a 'menu' button or something, and various system functions, load/save etc. you changed 'converter( )', assigned to the 'decimal' fader - didn't see that, i've been using the popup selector - because i have plans (har har har) to have the name of the TONE displayed, as assigned in the Timbre, as stored in these eg: ... fetch all Timbre tables, get toneGrp/toneNmber for each Timbre, fetch name (10 values) of each tone used, display this next to each Timbre number. this will give a popup of all timbres, with the names of the tones used in them. that's how the D-110 displays Timbres assigned to Parts, showing the name of the Tone used by the Timbre. Timbres don't have names, only A11-88,B11-88 numbering. Tones have names, and Patches have names. very snappy and straightforward, how you do the retrieval of values. but it needs the 'write', to commit it to the memory slot
or the above is completely wrong lol but it's what it looked like it was doing last time i powered it up. |
Beta Was this translation helpful? Give feedback.
-
|
here is a rough view of what (i think) a Patch looks like:
for each column we see the 11 'Part parameters' (as they are called as part of the 'Patch'), with 'Timbre' as the subset of parameters that are given values by the Timbres loaded to them. the same 7 parameters are saved when you save a new Timbre memory. the others belong to the Patch: pan, level, keyLimitLO, keyLimitHI and here's the fun bit, when you save a Patch, it saves those 11 parameters per part. in the sysex pages: there are 7 Timbre parameters, 11 Part parameters, and 11 Timbre Temporary area parameters. sorry for the long roundabout explanation. thought it was worth clarifying. this is at odds with Sean Luke, author of EdiSyn, who says that 'timbres' are irrelevant and not used - they are not required to write a new Patch; the Patch will save new settings OR the settings and number of the Timbre assigned to a Part. at least this is my current conclusion, and happy to be contradicted. |
Beta Was this translation helpful? Give feedback.
-
|
oops : forgot two parameters per Patch Part: MIDI channel, and Partial Reserve. these are listed/defined as part of the Patch but are not in the 'Part Parameters' subset/addresses. 'Partial Reserve' follows consecutively from Patch Name and Reverb settings, as do MIDI Channel; then each 'Part' has own sub-address numbering: 00 1F, 00 2B, 00 37, .. etc. - eg: going up in 12s from 00 1F, with the sub-address/offset counting 00h-0Bh with 07h as dummy byte (after the 7 timbre parameters). also checked 'timbre temorary area' (5-2), and that ignores/has dummy bytes for 00 0C to 00 0F. otherwise same pattern for the other parameters. on p117, the chart for address headers, 'Tone Temporary Area' (5-1) shows each part's temporary tone having a different start address. http://cdn.roland.com/assets/media/pdf/D-110_OM.pdf and that's how the PG-10 controller accesses the Tone parameters on each part - not sure how unit#/basic channel operate here. (reviewing the overall D110 multitimbral structure for the nth time) |
Beta Was this translation helpful? Give feedback.
-
|
I think you mean Now you're getting into the weeds! Think of 04 00 00 as a location in memory on the D110: So in a sysex message the number starting at byte 6 is the memory location (04) for the data Each Tone can hold 246 bytes of data as you rightly calculated, Memory locations for the start of each of 8 tones are therefore: 0 | 0
Common parameters occupy the first 0E 14 bytes 00 - 0D so WG pitch Coarse for Tone No.1 Partial 1 will be F0 41 10 16 12 04 00 0E val cs F7 even though it says offset 00 in No. 5-1-2 in the manual. So at the start of each new table in the manual you need to start with the sum of the previous bytes in the previous table, but see next paragraph. It's not so easy!. But there are four partials each of 58 bytes 3A so: Common parameters occupy the first 0E (14) bytes 00~0D so WG pitch Coarse for Tone No.1 Partial 2 will be F0 41 10 16 12 04 00 48 val cs F7 even though it says offset 00 in No.5-1-2 in the manual.
Common parameters occupy the first 0E (14) bytes 00~0D so WG pitch Coarse for Tone No.2 Partial 1 will be F0 41 10 16 12 04 02 04 val cs F7 even though it says offset 00 in No.5-1-2 in the manual. So at the start of each new table in the manual you need to start with the sum of the previous bytes in the previous table, but see next paragraph. It's not so easy! But there are four partials each of 58 bytes 3A so Common parameters occupy the first 0E (14) bytes 00~0D so WG pitch Coarse for Tone No.2 Partial 2 will be F0 41 10 16 12 04 02 3E 48 val cs F7 even though it says offset 00 in No.5-1-2 in the manual. So for each partial you need to offset by 58 0x3A
It's really difficult to get one's head around but thinking of those numbers in terms of position in memory should help. You can either hard code these numbers into Ctrlr in the SYSEX field or calculate them on the fly using lua (possibly with the help of lookup tables). You could even just have one massive table with static lookups for each byte. It's probably still very efficient and you could basically just copy the manual into the table. I did this with the Yamaha TX81-Z panel I wrote. Have each modulator in a table as a key and have an offset address as a binary string if you like. Make the modulator name exactly the same as in the manual, so you don't go insane, adding an identifying trailing digit, or you could have a table of values that you multiply by the Tone number 1-8. Endless possibilities but as long as you take this top down approach you shouldn't get too lost as is so easily done with a complex machine like this! |
Beta Was this translation helpful? Give feedback.
-
|
hi, yeah sorry about stupid typo up there, 04 not 40. |
Beta Was this translation helpful? Give feedback.
-
|
This is the overall structure 😂 |
Beta Was this translation helpful? Give feedback.
-
|
revised view of what is contained in the Patch, and what isn't. ( this isn't the panel, it's just a quick visualisation, to see where i'm going with it). it could have a main view like this, reflecting more how the unit powers on - then you hit Timbre>Edit, and Edit again, to get into the Tone on the selected Part. so i guess i could have a uiTabs, with buttons flipping to the relevant Tone page.
btw (thanks for !) how did you calculate all those intervals in hex so quickly ?! i did it last time around but very laboriously, and it was very offputting contemplating how i'd deal with the parameter number count after Part1 (the easy one lol).
|
Beta Was this translation helpful? Give feedback.
-
That’s because I’ve done this several times and it’s essentially the same layout as the JD-990 😃 Once you realize 04 00 00 is the start address you can work it out from there. You don’t have to do it in hex either; you can use decimal in your coding and then maybe convert to MemoryBlock Hex string to send sysex; you could even have a lookup table of 1968 values! |
Beta Was this translation helpful? Give feedback.
-
|
ah JD990 too? they came to their senses afterwards with the JV series, if you look at the organisation of the sysex :-) as far as i recall: it doesn't have breaks at weird hex numbers :D and once again, trying to get into the head of the roland engineers, and how they got it all to work in their firmware ... |
Beta Was this translation helpful? Give feedback.
-
|
Here is a panel that accesses the correct (hopefully) sysex for each temporary memory byte. This could be a good starting point for a panel. Instead of creating 100s of modulators. I use two combos so you can check how the sysex would be generated for any modulator. This is kind of a top down approach. All you need to do now is assign modulators and name them according to the names in Edit 8/2023 Corrected a few errors |
Beta Was this translation helpful? Give feedback.
-
|
btw 'Partial Mute' is a tricky one (0Ch) because it's one value 00-0F for all partial mute button configurations - which we covered with the 'partial ops' method - which dealt with the LEDs. it's one of the 'common' parameters anyway. (but means that there are more modulators than parameters - thought about this whilst considering 'VST Index', which i'm ignoring for now because it's giving me crazy high numbers having had several panels open simultaneously .. (can VST Index numbering be done afterwards? 'automatically'/sequentially with a script?) |
Beta Was this translation helpful? Give feedback.
-
|
Yes, you can reassign vst numbers with a script. But that should be done towards the end of creating the panel. Here is an example: t= {
"my Modulator 1",
"my Modulator 400",
}
--
-- Called when a modulator value changes
-- @mod http://ctrlr.org/api/class_ctrlr_modulator.html
-- @value new numeric value of the modulator
--
generate_vst_numbers = function(--[[ CtrlrModulator --]] mod --[[ number --]], value --[[ number --]], source)
if myPanel:getBootstrapState() then
return
end
if source ~= 4 then
return
end
local no = panel:getNumModulators()
for i = 0, no - 1 do
--CLEAR ALL FIRST
mod = panel:getModulatorByIndex(i)
mod:setPropertyString("vstIndex", "")
mod:setPropertyString("modulatorVstExported", "0")
console("clearing " .. L(mod:getName()))
end
for i, v in ipairs(t) do
local n = panel:getModulatorByName(v)
n:setPropertyString("vstIndex", tostring(i))
n:setPropertyString("modulatorVstExported", "1")
end
end --function
--------------------------------------------------------
|
Beta Was this translation helpful? Give feedback.
-
|
oh thanks for that! and hahaa i saw what you did with layers in your last panel, putting the tabs as a dummy object in the background layer - it doesn't contain anything, just changes the background colours. -however i noticed 'called when the current tab changes' associated with the uiTabs triggers the method 'recordCurrentTab', which runs 'calculateSysex' method, and this SHOULD trigger updating of the sysex message - since it's the same method used by the two dropdown combos - but it isn't doing it - ? please explain this again, the global variable you create with: nb: here i come across the awkwardness of layers in some situations, where you have to manually mute layers to access those beneath. good to get a fresh take on that. so far i've only used simple Ctrlr graphics, avoiding importing much external resources. you could go crazy with that, and layers, transparency, etc. - in fact the only time i've used it was when i was getting the background to change colour for each partial, so your use of tabs like this is an alternative take on that - cool idea! with a lot of other possibilities. (seems you can access lower layers as long as the control is not obscured by something in front of it - yes i messed up my previous panel pretty good overdoing that with 'setVisible' lol, and when you have the panel in 'edit' state it means seeing everything at once - i had buttons to call the views separately though ) |
Beta Was this translation helpful? Give feedback.
-
|
ok i see for the global variable now. |
Beta Was this translation helpful? Give feedback.
-
|
Oh! That's because somehow the panel I posted doesn't contain the necessary code! Sorry - try this one: Edit 8/2023 Corrected a few errors |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
2nd chat with it just now, and i thought i'd show it some verbose crude methods and see if it could condense them, and if they'd still work, pasted over and compiled, into an existing panel. the first one was too long :D and it wouldn't have it. so i tried another one, and this time it gave a really short concise alternative. i put it in place of the original method, and it didn't crash, and seems to be working.
quick explanation of what it is: it's a continuous control, that selects which of 2 views is shown, for 2 lanes. there are 13 possible values.
i'll show what chatGPT wrote first, and then the original:
and here's mine:
(fingers crossed this all displays correctly when it hit Post)
Beta Was this translation helpful? Give feedback.
All reactions