Bug? i think so...multichannel in main.cpp

[stretch911]

I've been playing w/ several chips, heltec v2, mini esp32 wroom trying to get atomlight to work...
star effects still don't work for me (see other post)...
but my main issue was i couldn't get multichannel to work....atomlight is set for 4 channels at 53 pixels.
started suspecting the logic in main.ccp and seems like a copy and paste issue....LEDPIN0 in all conditions...
gonna fix my version and if it fixes will check in to repo.

#if USESTRIP

    #if NUM_CHANNELS == 1
        debugI("Adding %d LEDs to FastLED.", g_aptrDevices[0]->GetLEDCount());

        FastLED.addLeds<WS2812B, LED_PIN0, COLOR_ORDER>(g_aptrDevices[0]->leds, g_aptrDevices[0]->GetLEDCount());
        //FastLED.setMaxRefreshRate(100, false);
        pinMode(LED_PIN0, OUTPUT);
    #endif

    #if NUM_CHANNELS >= 2
        pinMode(LED_PIN0, OUTPUT);
        FastLED.addLeds<WS2812B, LED_PIN0, COLOR_ORDER>(g_aptrDevices[1]->leds,g_aptrDevices[0]->GetLEDCount());

        pinMode(LED_PIN1, OUTPUT);
        FastLED.addLeds<WS2812B, LED_PIN0, COLOR_ORDER>(g_aptrDevices[1]->leds,g_aptrDevices[1]->GetLEDCount());
    #endif

    #if NUM_CHANNELS >= 3
        pinMode(LED_PIN2, OUTPUT);
        FastLED.addLeds<WS2812B, LED_PIN0, COLOR_ORDER>(g_aptrDevices[2]->leds,g_aptrDevices[2]->GetLEDCount());
    #endif

    #if NUM_CHANNELS >= 4
        pinMode(LED_PIN3, OUTPUT);
        FastLED.addLeds<WS2812B, LED_PIN0, COLOR_ORDER>(g_aptrDevices[3]->leds,g_aptrDevices[3]->GetLEDCount());
    #endif

[stretch911]

seems to have fixed my multichannel issue. starting w/ 2 strips/spokes...will increment.
also, channel 2, pixel index [0] is not updating...will sniff this one out...
btw... @robertlipe.... famous geocache hider in Nashville. Has two of my most favorite...Key to Crytonomicon and the Morse code using low power microchip. small world.

[stretch911]

I checked those main.cpp fixes in. hope i did that right. created a fork? and pushed...

[robertlipe]

Nothing authoritative on your actual problem, but it sure *looks* wrong and I'd try really hard to make it more immune to copy paste issues. Try one or more of these: .. 1) put the LEDs into any array and index from numchan. 2l Use if constexpr() instead of ifdefs to keep unwanted ifs out of the optimized binaries but still compiling. 3) avoid code repeats and reuse a set variable (probably computed in #1) and reuse the setpin, address, and index of Devices[]. 4) consider putting the array in a hash map so it's trivial to check for uniqueness (it's not exactly hard even with C arrays with small sizes like this...) at population time to be sure that the user didn't misconfigure the pin #defines. When you're done,. It'll hopefully work the same, but be a little smaller, and, more importantly, more immune to future changes. (This might technically run afoul of the blinkenperbit metric, but I'd hope there is space for a negative metric score that improves engineering taste...) If there's a stomach for such changes and you can help me build up a repro setup with a minimal collection of strips and you'd like some mentoring/partnership opportunities, I'm open to help coach and/or slinging some of those bits. I probably have at least four 2812 strips, but they're all of different lengths and densities. (Do we have food standalone unit tests for such things without hardware?) If I can use some of the 2812 (not HUB75) panels and do true/false testing with a few cells per hose, I think I can wire that up. Do we allow strips of different types? Can strip 0 be 5v 2812s and strip 1 be 12v 2811s and strip 2 be apa104s or other masochistic configurations? I'm a big fan of making software hard to misuse by end-users and developers. LMK if you're interested in tackling this together. And, yes, I am, or rather was in my able-bodied days, that same catcher known for PITA placements and being ranked as high as #6 in the world for a while, back in the very early part of the century. I helped create the world largest annual gathering of geocachers (GeoWoodstock) and the software used by almost every other geocaching software for interfacing with GPS hardware and common map programs, GPSBabel. Thanx for the shout-out!

…

On Wed, Jun 21, 2023, 9:28 AM stretch911 ***@***.***> wrote: I checked those main.cpp fixes in. hope i did that right. created a fork? and pushed... — Reply to this email directly, view it on GitHub <#328 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACCSD3ZZSFXG2WWIMVSN6K3XMMAJNANCNFSM6AAAAAAZOXJ64I> . You are receiving this because you were mentioned.Message ID: <PlummersSoftwareLLC/NightDriverStrip/repo-discussions/328/comments/6242684 @github.com>

[stretch911]

i actually am up for some mentorship. current profession is automating/performance testing applications. i use java/groovy/python/selenium/jmeter for my tools.

you are LOCAL too for me. I'm in White House (think you're in Franklin area).
I have some left over 2812s 60 per meter from my kitchen led project.
I built an atomlight several months ago, but still can't get star effects to work and the code is so deep i can't follow to debug. plus i'm not a c++ guy but have programmed in c so between all the above languages it doesn't look awefully difficult. i fixed the multichannel thing but would like to know/do what you said ....i'll have to reread and absorb.

glad you didn't mind me shouting you out. the microchip geocache multi on the hill in franklin was my 200th cache. and key to crytonomicon my 500th. i have a few favs in franklin too! but nothing w/ microchips!!!!

[stretch911]

Do we allow strips of different types? Can strip 0 be 5v 2812s and strip 1
be 12v 2811s and strip 2 be apa104s or other masochistic configurations?

no...i have used the leds w/ the white channel...but effects look aweful. took me a minute to figure out i had the wrong strip hooked up and then i was amazed at the sexy effects from dave's programming

[robertlipe]

The hardcoded WS2812 in those very lines where they've been added to FastLED is probably the ("a") place to change that if we want to revisit that. So for now, we'll assume the world is WS2812's (or compatibles) and march on ... after we hear from a reviewer that this is a sensible direction to go. So let's get a head-nod of agreement that this is a sensible thing to "overfix" (beyond just fixing the obvious typos) before we unholster our keyboards. How much of what I suggested are you comfortable slinging? I can review or work through it together or I can just write it and touch test it and let you do the final test. Your call. My gmail address is the same as my playername. I was in Franklin for 25+ years. Brentwood/Antioch now. White House is still closer than Shenzhen, so LMK what kind of trading material you're looking for. I have some spare perf boards or LCD panels and such. I can find a home for some short 2812 strips and/or the 3-pin JST-SM->Bare wires for pairing my various strips,reels, and grids into the MCUs since it didn't occur to me that those wouldn't be included in most cases. But I think I can at least get a run on this project with what I have; dupont wires to the rescue. This <https://www.youtube.com/watch?v=_wCOCI18nAk&ab_channel=Dave%27sGarage> one? I remember noting it before I latched on here because it reminded me of the light <https://www.amazon.com/gp/product/B06X92R1XH/> I installed in my dining room. Just speeding through it, it looks like most "magic" is done in the software and that's presumably already written and debugged. So if something is wonky, I'd assume electrical issues. But if this isn't your forte and you're in no hurry, I can bring it to Lipe Labs and stick it on meters and scopes and trace wires with fresh eyes and see if I can help awesomize it appropriately. I'm slow, but I'll take a look at it for you. The only (?) thing that looks easy to get out of sync would be mapping the four data lines to the pins on the ESP. If those are out of order, I can imagine it being non-symmetric in some weird ways as spokes that should be opposite might be next to each other or something. Then I'll get the joy of "creating" one without actually spending money and trying to find a flat space to store another object. :-) Buzz me at gmail and we'll separate the part of this discussion not related to "fixing" the bugs here to a private line.

…

On Wed, Jun 21, 2023 at 2:37 PM stretch911 ***@***.***> wrote: Do we allow strips of different types? Can strip 0 be 5v 2812s and strip 1 be 12v 2811s and strip 2 be apa104s or other masochistic configurations? no...i have used the leds w/ the white channel...but effects look aweful. took me a minute to figure out i had the wrong strip hooked up and then i was amazed at the sexy effects from dave's programming — Reply to this email directly, view it on GitHub <#328 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACCSD35FMVYGS3Z2GSQCY7LXMNEP7ANCNFSM6AAAAAAZOXJ64I> . You are receiving this because you were mentioned.Message ID: <PlummersSoftwareLLC/NightDriverStrip/repo-discussions/328/comments/6245391 @github.com>

[davepl]

The current model has been to write a specialization of GFXBase. So there’s LEDMatrixGFX, LEDStripGFX, and HexagonGFX, for example. What I might suggest would be to add an “InitializeHardware()” virtual function to GFXBase. Then LEDMatrixGFX could do its hardware-specific init within that function. LEDStripGFX would call addLEDs in that function rather than in setup(). Perhaps LEDStripGFX should become LEDWS2812BGFX and implement the addLEDs speciic to the template, unless it can be done as a variable. That way, if someone wrote an impl for a different strip chip, they could just define their class similarly. Just a thought! - Dave

…

On Jun 21, 2023, at 1:54 PM, Robert Lipe ***@***.***> wrote: The hardcoded WS2812 in those very lines where they've been added to FastLED is probably the ("a") place to change that if we want to revisit that. So for now, we'll assume the world is WS2812's (or compatibles) and march on ... after we hear from a reviewer that this is a sensible direction to go. So let's get a head-nod of agreement that this is a sensible thing to "overfix" (beyond just fixing the obvious typos) before we unholster our keyboards. How much of what I suggested are you comfortable slinging? I can review or work through it together or I can just write it and touch test it and let you do the final test. Your call. My gmail address is the same as my playername. I was in Franklin for 25+ years. Brentwood/Antioch now. White House is still closer than Shenzhen, so LMK what kind of trading material you're looking for. I have some spare perf boards or LCD panels and such. I can find a home for some short 2812 strips and/or the 3-pin JST-SM->Bare wires for pairing my various strips,reels, and grids into the MCUs since it didn't occur to me that those wouldn't be included in most cases. But I think I can at least get a run on this project with what I have; dupont wires to the rescue. This <https://www.youtube.com/watch?v=_wCOCI18nAk&ab_channel=Dave%27sGarage> one? I remember noting it before I latched on here because it reminded me of the light <https://www.amazon.com/gp/product/B06X92R1XH/> I installed in my dining room. Just speeding through it, it looks like most "magic" is done in the software and that's presumably already written and debugged. So if something is wonky, I'd assume electrical issues. But if this isn't your forte and you're in no hurry, I can bring it to Lipe Labs and stick it on meters and scopes and trace wires with fresh eyes and see if I can help awesomize it appropriately. I'm slow, but I'll take a look at it for you. The only (?) thing that looks easy to get out of sync would be mapping the four data lines to the pins on the ESP. If those are out of order, I can imagine it being non-symmetric in some weird ways as spokes that should be opposite might be next to each other or something. Then I'll get the joy of "creating" one without actually spending money and trying to find a flat space to store another object. :-) Buzz me at gmail and we'll separate the part of this discussion not related to "fixing" the bugs here to a private line. On Wed, Jun 21, 2023 at 2:37 PM stretch911 ***@***.***> wrote: > Do we allow strips of different types? Can strip 0 be 5v 2812s and strip 1 > be 12v 2811s and strip 2 be apa104s or other masochistic configurations? > > no...i have used the leds w/ the white channel...but effects look aweful. > took me a minute to figure out i had the wrong strip hooked up and then i > was amazed at the sexy effects from dave's programming > > — > Reply to this email directly, view it on GitHub > <#328 (reply in thread)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/ACCSD35FMVYGS3Z2GSQCY7LXMNEP7ANCNFSM6AAAAAAZOXJ64I> > . > You are receiving this because you were mentioned.Message ID: > <PlummersSoftwareLLC/NightDriverStrip/repo-discussions/328/comments/6245391 > @github.com> > — Reply to this email directly, view it on GitHub <#328 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA4HCF5MWT2D6BPJV6RDJ2TXMNNQVANCNFSM6AAAAAAZOXJ64I>. You are receiving this because you are subscribed to this thread.