Comments by honey the codewitch

honey the codewitch 20-Jul-24 0:29am View

FreeRTOS has a built in thread safe circular buffer for uneven length items, but I'm not sure it's appropriate here.

honey the codewitch 26-Jun-24 14:24pm View

It makes a surprising amount of sense to me. It's simpler than anything I expected I'd have to do.

honey the codewitch 26-Jun-24 10:49am View

actually someone just solved it for me. Thanks!

and it was simple. so simple I feel dumb. LOL

// sets the integer channel value by name
template<typename name="">
constexpr inline void channel(typename channel_by_index<channel_index_by_name<name>::value>::int_type value) {
constexpr const int index = channel_index_by_name<name>::value;
channel<index>(value);
}

// sets the integer channel values by name
template<typename name1,="" typename...="" names="">
constexpr inline void channel(typename channel_by_index<type::channel_index_by_name<name1>::value>::int_type value1,
typename type::channel_by_index<channel_index_by_name<names>::value>::int_type... values) {
constexpr const int index1 = type::channel_index_by_name<name1>::value;
channel<index1>(value1);
channel<names...>(values...);
}

honey the codewitch 26-Jun-24 10:41am View

template<typename name="">
constexpr inline void channel(typename channel_by_index<channel_index_by_name<name>::value>::int_type value) {
constexpr const int index = channel_index_by_name<name>::value;
channel<index>(value);
}

and then

template<typename name=""> using channel_index_by_name = typename helpers::channel_index_by_name_impl<0,Name,ChannelTraits...>;

honey the codewitch 26-Jun-24 10:23am View

I'll have to figure out SFINAE with std to do that.

honey the codewitch 26-Jun-24 10:23am View

The problem isn't there. It's when I actually go to use it, I ended up needing two. One for the names, and one to get the info for each channel so I can actually set it.

honey the codewitch 24-Jun-24 5:37am View

Thanks. Looking now.

honey the codewitch 24-Jun-24 5:11am View

I'd love to tie my graphics library to ARM, but alas, it would greatly narrow its use for IoT and hobbyist endeavors which is probably where other people use it the most. I wrote it for use in professional projects because I needed a lib I could add features to. Plus when I wrote it, Mine was the only lib to support TTF in under 192KB of SRAM total system memory. (I don't know how little it requires actually, and it kind of depends on the font). It also supports SVG, and is the only embedded lib to do so, to my knowledge.

honey the codewitch 24-Jun-24 5:03am View

hazards of embedded, if I use std::tuple<> I've got to enumerate all the embedded platforms that don't support it properly so I can make it work for those. 8 and 16 bit AVRs are a good example of a platform whose STL implementation is incomplete and/or non-compliant. Not using it is a lot less work.

honey the codewitch 24-Jun-24 4:46am View

Interesting. I did not know it was possible to produce a parameter pack in these versions of C++. That's what I need. I can't use std::tuple but it's using C++ language features so I just need to figure it out.

honey the codewitch 23-Jun-24 22:06pm View

Oh and as far as me representing pixels as bitfield structs, that's not practical here for a number of reasons.

honey the codewitch 23-Jun-24 22:04pm View

unfortunately, there's no such corollary with values... here in C++ as the same name is essentially applied to every argument.

honey the codewitch 12-Jun-24 5:44am View

Thank you. I appreciate the input

honey the codewitch 11-Jun-24 23:44pm View

I think you're right. I'll fix it now. It slipped past me because the failure doesn't actually cause anything visible to not work - just a portion of the screen to refresh more frequently.

honey the codewitch 3-Apr-24 5:59am View

Thanks!

honey the codewitch 3-Apr-24 5:23am View

The problem with React (and maybe I'm wrong) is it requires server side. I need strict HTML/CSS/JS. That means no typescript, no server transformations, etc.

I might be stuck with JQuery.

honey the codewitch 8-Feb-24 15:07pm View

for the record, it wasn't me that downvoted you. I haven't voted on any of the solutions.

honey the codewitch 8-Feb-24 11:22am View

That's where i'm at right now, but I can't shake the feeling of being on borrowed time.

honey the codewitch 8-Feb-24 9:21am View

Thank you. That gives me plenty of food for thought.

honey the codewitch 8-Feb-24 4:40am View

Thanks! I'll give that a look

honey the codewitch 8-Feb-24 4:31am View

Fair enough. I'm not really wedded to UWP, i just thought that was the next jump off point since WinForms is going the way of the dodo. What do you use for writing desktop apps in .NET these days?

honey the codewitch 8-Feb-24 4:06am View

what did you end up doing about it?

honey the codewitch 8-Feb-24 3:49am View

Thank you. I'll bookmark that, but I've also just had AvaloniaUI and Blazor recommended as potential alternatives. Looks like I have a lot of legwork to do, but this is a much better position to be in than I was in when I asked initially. =)

honey the codewitch 8-Feb-24 2:55am View

Yeah. To people that use UWP, what does your effing process look like?

honey the codewitch 8-Feb-24 2:54am View

Haha, I put it here because I thought it was too much of a programming question to be in the lounge. In fact, a may just cross post there now so I can get yelled at in both places. =)

Truth is, I *am* stuck. I can't use these tools as provided by Microsoft. There has to be a way people work with UWP and the like every day, but as it's packaged and sold? This technology is cumbersome, confusing, and near impossible to use fluidly. I do need help.

honey the codewitch 5-Jan-24 12:53pm View

There's probably a way but I don't know it.

honey the codewitch 5-Jan-24 11:41am View

Thanks for the suggestion. I just tried that. I get nothing. In fact, it runs the code after that line, so I don't think it's throwing anything.

honey the codewitch 11-Nov-23 20:14pm View

I solved it but I'm not sure how exactly so I don't know what to put for a solution!

honey the codewitch 11-Nov-23 17:13pm View

its usb cdc serial. if there's a way to control handshaking on it, i wouldn't know. this same technique works for other projects.

honey the codewitch 9-Nov-23 9:15am View

In that case it's not practical, because the user may need to select the heap to use at runtime, and will typically be using multiple heaps at the same time. Overloading new *could* get you that, if you made it call out to some switchable allocation scheme, but it's clumsier than using emplace new inside my libs

honey the codewitch 9-Nov-23 8:55am View

In IoT and embedded you have different physical heaps with different sizes and performance characteristics. You have PSRAM and SRAM for example. Because of this, a user of my lib can choose the allocators to use when it needs to dynamically allocate RAM. Right now I've been very careful about RAII on anything that is the target of those allocations, but I'd rather use placement new and fire the constructors cleanly.

honey the codewitch 9-Nov-23 8:53am View

+5 thanks. Wow, here I thought it was just a variant of the keyword new. I guess it's a function? Odd.

honey the codewitch 9-Nov-23 4:36am View

This is the standard new operator. Not the emplace new operator. I am trying to call new on an existing buffer using the emplace new operator.

honey the codewitch 4-Oct-23 9:49am View

didn't work

honey the codewitch 4-Oct-23 9:43am View

Ah, I'll give it a shot

honey the codewitch 30-Sep-23 0:30am View

It didn't seem to make any difference, as long as I was consistent (always using the dllmain hinstance vs always using the result of GetModuleHandle(NULL) ). They were clearly different handles, but windows didn't seem to care. I imagine it makes a difference if you were loading resources though, because if you used the GetModuleHandle(NULL) it would try to load them from the exe instead of the DLL. Since I'm not using resources it made - as far as I can tell, no discernable difference.

honey the codewitch 24-Sep-23 7:07am View

I've tried both. Same result.

honey the codewitch 23-Sep-23 22:25pm View

if you're game.. https://github.com/codewitch-honey-crisis/winduino/blob/main/winduino_hardware/spi_screen/spi_screen.cpp

Start scanning after line 240, see if anything jumps out at you. And pardon the mess. There is no factoring right now. Just trying to get it to work as a proof of concept at the moment.

honey the codewitch 23-Sep-23 22:06pm View

I was able to get it to start logging, and it has called my WndProc, so that's not the issue.

The other thing that's weird, and that I don't understand at all, is that window is the ONLY one that's hourglassing and going non-responsive. This despite the fact that all windows are on the same thread/message pump. How is that even possible? They are cooperatively tasked within the same thread.

honey the codewitch 23-Sep-23 21:12pm View

He's saying their shouldn't be any problem since everything shares the same process address space, as long as it's all operating on the same thread (which it is).

All of that is what I expect. I was looking for things I *wasn't* expecting at this point.

I still can't find the problem, but I don't think it's the above. I just don't know what it is.

honey the codewitch 23-Sep-23 19:13pm View

I'll go ahead and accept this, as it all jibes with what I was expecting. Seems my problem lies elsewhere.

honey the codewitch 23-Sep-23 18:37pm View

Yeah, I'm aware of what an hinstance is. But I did not know about that bug, thanks. I tried replacing GetModuleHandle(NULL) with the hinstance from DllMain and it did not like that at all. It wouldn't even register the window class i think (hard to tell because i can't debug and logging is limited)

honey the codewitch 23-Sep-23 16:31pm View

I'll try that. I'm using GCC and don't currently have a DllMain to grab the HINSTANCE off of but I can sort that out, thanks.

honey the codewitch 22-Sep-23 7:01am View

thank you again!

honey the codewitch 22-Sep-23 7:00am View

Should be, but aren't, and I don't care, because the other HWND works for SetWindowText. Good enough for me. GDB doesn't like me anyway.

honey the codewitch 21-Sep-23 22:00pm View

I was hoping you could clarify something since I'm not seeing the sample you're suggesting.

The following is in my code. This appears to be user mode, because _KERNEL_MODE is not defined in my code. So am I already using the User Mode Driver Framework here?

#ifdef _KERNEL_MODE
#include <ntddk.h>
#else
#include <windows.h>
#endif

#include <wdf.h>

honey the codewitch 21-Sep-23 3:55am View

Amazing! Thank you so much!

honey the codewitch 20-Sep-23 15:08pm View

thanks. i'll give it a look

honey the codewitch 20-Sep-23 14:07pm View

That's what I've started with, but I can't find that. I'll look again.

honey the codewitch 13-Sep-23 12:18pm View

My code above was fine for the most part. The issue was my graphics library! There was a latent bug, sitting in there for maybe a year or more? It didn't handle bitmaps with alpha channels correctly - i was actually corner casing for it when I shouldn't have been. Simplified the code to remove that special casing, and bam! everything works as expected. Thanks again!

honey the codewitch 13-Sep-23 11:21am View

I had to do some porting because I'm on GCC but I'm going to go ahead and accept this. Copying your code into mine didn't help, so my problem lies elsewhere. I will instead start with your code, and build it up. Thank you for your help.

honey the codewitch 13-Sep-23 10:39am View

In my code, any pixel created with an alpha channel, such as the bgra_pixel<32> used above has it's alpha defaulted to full opacity, in this case 0xFF at 8-bits per channel.

gfx::draw::filled_rectangle(bmp,bmp.bounds(),color_type::purple);

This line should fill m_frame_buffer appropriately with like 0xFF 0x00 0xFF 0xFF like that. (or similar, I actually don't know what the X11 values for purple are off the top of my head)

honey the codewitch 13-Sep-23 10:26am View

Oh I'm calling InvalidateRect in my message loop for now.

honey the codewitch 13-Sep-23 9:58am View

Thanks. I'll give it a shot. I meant Direct2D, and yet always type DirectDraw because I learned DirectX ages ago, and the old tech stuck. I hardly do any windows development at all. Mostly embedded so I don't have much reason to try to retrain myself.

honey the codewitch 29-Aug-23 9:03am View

sorry! didn't know

honey the codewitch 25-Aug-23 7:44am View

I have not. I moved on but I still want to find the problem.

honey the codewitch 21-Aug-23 16:38pm View

that's a good idea. maybe i'll give that shot, though it's a bit involved making one of those fonts from scratch. Perhaps i can modify one.

honey the codewitch 21-Aug-23 14:56pm View

m_current is const uint8_t*

honey the codewitch 21-Aug-23 14:56pm View

the seek_buffer is similarly a thin wrapper. I'm confident in the seek code since I use it extensively for rendering straight out of TrueType font files, in maybe hundred projects now. It's rock solid. I think. :~

honey the codewitch 21-Aug-23 11:50am View

Yes! My terrible self and a bad edit. (I removed some code to post here because it was in the way)

honey the codewitch 21-Aug-23 8:40am View

I am indeed but it's just this: *in_out_size=stm->read(buffer,*in_out_size); return gfx_result::success;

Edit: I think I just found the bug!
Edit 2: Nope.

honey the codewitch 21-Aug-23 8:26am View

yeah. It's a simple delegation used to just workaround the fact that my containing class can be const, but a stream read is not const because it changes the cursor position. So I forward to a static method, passing the stream, effectively circumventing the const restriction. It's sort of kludgy, but it's due to a const "polarity" mismatch introduced by the fact that the stream has a cursor. It's just two lines: *in_out_size=stm->read(buffer,*in_out_size); return gfx_result::success;

honey the codewitch 21-Aug-23 8:22am View

Oh, I'm probably not going to use your workaround, no offense! :). I'm going to find the problem. I can't have this code underneath read_stream() not working properly, as it's mission critical in a lot of projects.

honey the codewitch 21-Aug-23 7:16am View

Thanks. Exactly what I needed before I went further. Everything you say jibes with what I understand, but the stream code itself is used in professional projects. Yikes. It's never broken before that I know of, and I use it for complicated stuff like streaming TrueType off of a file

honey the codewitch 21-Aug-23 1:08am View

My bitmap code is such that there is no end of line padding, but yeah I know what you mean. I have to do some acrobatics on an e-paper panel driver i wrote due to that.

honey the codewitch 21-Aug-23 1:07am View

read_data simply reads from the stream at the last place it was seeked to, and advances the cursor by however many bytes were actually read.

honey the codewitch 23-Jul-23 7:59am View

Ah thank you. I must have been looking at docs for the old standard. I had been at this stuff for so long at that point yesterday it doesn't surprise me.

honey the codewitch 23-Jul-23 3:50am View

For reasons known only to K&R NAN and INFINITY are not defined by C. Anywhere. Sometimes, some implementations of math.h provide it, but not on the systems I'm targeting.

Furthermore, while my graphics library does not rely on the STL it is in fact C++14 or C++17 (depending on the headers used). I know people use C++ with ZephyrOS, so maybe this is a zephyrOS question.

Thanks for your response. Basically it confirms for me that this is most likely a GCC vs G++ issue, and not a problem finding includes.

honey the codewitch 8-Jul-23 11:33am View

I've added some code in What You Tried. I think gets me really close. There may be ways I can optimize it though.

honey the codewitch 8-Jul-23 9:24am View

The solution I need is actually a lot more complicated than what you described. I really should make something in MSPaint to visualize it but it would be rather time consuming and labor intensive. The issue comes from filling around the existing rects with irregular rects. It's The Tiling Problem, but with irregular tiles (not on the grid) - in fact, the Tiling Algorithm may work if I make the grid 1x1 but it would be inefficient.

Edit: I've added code in What You've Tried that works, but may be made more optimal.

honey the codewitch 8-Jul-23 9:20am View

your comment got cut off. Maybe "improve" your existing solution with the code. And thanks again. This is getting interesting.

honey the codewitch 7-Jul-23 14:42pm View

It can allow overlaps in theory, it's just suboptimal as it will draw to the display twice in some areas when once would have done. I can't use Win GDI/+ calls because this is not tied to windows. In fact the codebase is primarily meant for little IoT widgets. It *will* run on a PC, but there's little sense in doing it except for prototyping. That said, my graphics library is, while not quite as full featured as GDI+, it's at least in spitting distance in a lot of ways. The API is very rich, and I can take advantage of it.

honey the codewitch 7-Jul-23 14:32pm View

there's effectively no minimum but since the idea is to minimize the number of rectangles needed, each rectangle needs to be as big as possible, limited to 32KB in area.

honey the codewitch 7-Jul-23 14:31pm View

To be clear the pre-drawn untouchable rectangles may be larger than 128x128 in area. It's just fill rectangles where I'm limited to that - the ones i have to create. Other than that yes, that is my issue. Like I said I looked at the tile problem demonstration and description and it doesn't quite fit my problem. Almost, but because it doesn't a different algorithm is needed. Someone suggested a kd-tree might be useful for this, but i don't know. I've only ever used that animal for nearest color matching.

honey the codewitch 7-Jul-23 14:28pm View

The algorithm isn't sufficient. The reason that I didn't know the name of it is probably because I never went to school for any of this. Being self-taught, there are holes in my knowledge large enough to drive a truck through. Oh well.

honey the codewitch 7-Jul-23 9:37am View

Thank you for pointing me to the algo. After looking it up: https://www.geeksforgeeks.org/tiling-problem-using-divide-and-conquer-algorithm/ it seems like it almost describes my problem but not quite, because the filled in tiles occupy locations that are regular in size and location - they fit in the grid. My existing rectangles do not fit in a grid, so the algorithm I linked to won't work. I could subdivide into smaller grids so that the existing rects take up multiple grid spaces, but the problem with that is the grid is very small and I'd need to recombine small rects into larger rects, which is a problem in and of itself. Once again, thank you for pointing me to this algorithm, even if it's not quite what I need.

honey the codewitch 7-Jul-23 9:24am View

It's not quite what I need. The issue is it that it needs to fill in irregular sections that aren't 128x128, but may be 32x200 for example. In your code you simply don't put in the rect if it overlaps. This actually gets me my first pass so your solution is partially correct. Basically what I need is another pass to go through and fill in the blank spots. However, this still won't yield optimal results, but it gets me closer.

honey the codewitch 7-Jul-23 9:18am View

I'm setting it up in C++ now, with visualization for checking. I'll let you know how it goes. It may be a bit, as I'm still waking up, drinking coffee, and poking at it periodically at the moment. Thanks again. :)

honey the codewitch 7-Jul-23 8:52am View

I'm plus +5 ing this for the sheer effort. It may solve my problem but I need to try it. It certainly looks like it works. It will take a minute to test everything so I am accepting this solution in the interim to keep people from posting more solutions. I may reject it after-the-fact if it doesn't solve the problem. No matter what happens, thank you very much for this.

honey the codewitch 7-Jul-23 8:49am View

Yeah. Only thing is the rect locations aren't technically random, but they are for this purpose. IRL they are laid out controls.

honey the codewitch 6-Jul-23 9:59am View

yeah. sorry. all the equality types in JS make me cross-eyed. ha!

honey the codewitch 6-Jul-23 9:58am View

I think it is! I just couldn't see it. I'll test it, but mark this as solved, because I'm 80% sure you just unstuck me.

Actually I think it's: return (m_buffer_size * 8 + 7) / typename bitmap_type::pixel_type::bit_depth;

But close enough! The purpose of +7 is basically the same as ceil() on the result of the division, were it floating point. That way 9 px at a bit depth of 1 takes 2 byte, not 1 byte. :)

Edit: Wait. I think this is wrong. meh.

Edit 2: No it's right.

honey the codewitch 6-Jul-23 9:55am View

Is that JS? The triple equals is throwing me off. I'm assuming it's the same as == in C++

honey the codewitch 27-Jun-23 11:42am View

This is the only place new or delete are being called, in the shown code. this is on ESP32s and SAMD51s, not ARMs. I don't have a debugger on these devices. Even JTAG is too slow to be usable. Besides, I'm not getting a stack dump. I'm getting a hang. And only on devices with a certain type of LCD display bus.

honey the codewitch 26-Jun-23 20:03pm View

Those copy constructors are never getting invoked in an instance where it would create an invalid linked list. That much is tested code. So the delete code works correctly. Every next pointer created is valid until segment_clear_all() at which point none are.

honey the codewitch 22-Jun-23 19:42pm View

It kind of depends on what you mean by "copy". Both ways are valid C++, but the kind of copy it is is a little different. If I copy a segment that's linked to segment B, then there will be two segments leading to segment B. That's an invalid link list of course, even though the copied instance is valid which owes to your concern, but the way this copy constructor is used that will never be an issue. Good eye though. :)

honey the codewitch 22-Jun-23 14:27pm View

thanks. yeah i could put that in to make it more explicit but it's essentially already checked by other if.

honey the codewitch 4-May-23 10:01am View

Yes. I run make clean before I build. I got it working using an older version of the actual toolchain - not changing the build scripts at all. I still don't know why though, which is why I'm leaving this open.

honey the codewitch 28-Apr-23 7:20am View

It's possible, but I have some difficulty assuming that because xboot is widely used for AllWinner chips (at least by Chinese folks), and I've had a heck of a time running down information on using it. Hard to search google in Chinese, at least for me. I'm going to try this on my VM again. I remember it worked there. It's less than ideal, but maybe it will give me some insight if i make with -d and compare the results. Gonna do it once I feel up to tackling it.

honey the codewitch 27-Apr-23 12:53pm View

That's a good question. I assumed it was omitting something but I think you're right, that's the full path. Now I'm wondering why it worked before. I may have to unpack xboot again

honey the codewitch 27-Apr-23 10:17am View

i just added the final part of the output to the question

honey the codewitch 27-Apr-23 8:55am View

I used strace and I can't figure out what this line means
make[2]: /home/honey/arm-gnueabihf-gcc: Permission denied

Is it trying to execute the directory?

honey the codewitch 27-Apr-23 7:43am View

I've tried strace, and I don't remember the result because it was when I first ran into this problem. I'll try again

honey the codewitch 27-Apr-23 7:43am View

Here's a sample. It all looks like this (there's a lot of it)
-rwxr-xr-x 1 honey honey 10392 Jul 1 2021 FrameDecorator.py
-rwxr-xr-x 1 honey honey 1577 Jul 1 2021 FrameIterator.py
-rwxr-xr-x 1 honey honey 6758 Jul 1 2021 __init__.py
drwxr-xr-x 2 honey honey 4096 Jul 1 2021 command
-rwxr-xr-x 1 honey honey 8031 Jul 1 2021 frames.py
drwxr-xr-x 2 honey honey 4096 Jul 1 2021 function
drwxr-xr-x 2 honey honey 4096 Jul 1 2021 printer
-rwxr-xr-x 1 honey honey 10913 Jul 1 2021 printing.py
-rwxr-xr-x 1 honey honey 4209 Jul 1 2021 prompt.py
-rwxr-xr-x 1 honey honey 5530 Jul 1 2021 types.py
-rwxr-xr-x 1 honey honey 3137 Jul 1 2021 unwinder.py
-rwxr-xr-x 1 honey honey 11077 Jul 1 2021 xmethod.py

honey the codewitch 22-Apr-23 11:27am View

That could very well be, but we also can't offer the things a lot of other outfits can. We're kind of a shoestring outfit. I'm not sure how much we could raise our rates and remain competitive.

honey the codewitch 21-Apr-23 16:03pm View

It varies, but we're trying to keep it flexible, so the idea is compact, modular, and relatively inexpensive. How compact? Dunno yet. How cheap? < $50 to manufacture the main "brain" (not including the modules), another $20-$40 for a screen. None of these prices are bulk. That way we don't have to make a ton of different devices to meet a bunch of needs. Based on the projects I've done this main brain can handle anything I've had to do for a client by a factor of 10x. I need 100mhz. I have 1Ghz. I need 512KB RAM. I have 512MB. The only thing I'm kind of lacking is flash space, but I could use SD internally for gigs of space, which i may do if I can make it reliable.

honey the codewitch 21-Apr-23 12:37pm View

It's not the cost exactly, but the payment structures that concerns me. I charge for deliverables, so I don't want to push recurring costs onto my clients unless they have a specific service need for their project.

As far as chips being swapped that's the problem. These ARM based SoCs are so diverse every single one is going to have different registers, if not a different footprint because they have different on-chip peripherals.

I think I've found a solution in the allwinner H616 even though it's meant for TV boxes like making a Roku knock off. It is Cortex A53 based w/ HDMI which is what I wanted.

Here are my notes on this solution. If you want to take a look and offer your input I wouldn't mind, but no obligation

https://pastebin.com/4p0HWAKB

honey the codewitch 12-Mar-23 12:10pm View

This is a dual core 240MHz proc with about 360KB of usable SRAM. I am using TTF because frankly, it's the quickest and dirtiest way to get anti-aliased fonts onto my screen using the libraries I'm using, and gives me the option to scale to different displays if I want to support different hardware. If I was super concerned about performance in this instance, I could prerender to a bitmap, but you run into perf issues anyway due to blending required for kerning.

Edit: In my LVGL TTF lib I do glyph caching so what you're talking about happens in the background, but I haven't implemented that for htcw_gfx yet.

honey the codewitch 12-Mar-23 11:44am View

You're leaning me toward that option. The screen isn't very wide or I'd have considered doing what you say, but I have maybe a dozen characters left to right - i haven't counted yet, or rescaled the font or anything just yet. My app in fact, computes the max column count based on the size of the display.

honey the codewitch 22-Feb-23 14:17pm View

As best as I could visually verify yes. I did similar. I'm closing this solution as solved because I'm near certain the bug is coming from my rendering routines. Maybe deep in my graphics library, brought about because of negative coordinates off the top of the bitmap. It *should* clip, and draw appropriately but it seems it's not. I think it might be due to converting something from signed to unsigned without cropping it first.

honey the codewitch 22-Feb-23 10:14am View

I'm keeping this solution open pending further investigation because I've tried it and it's not working but now I think the bug is probably somewhere else. Thanks for finding that 1 pixel bug =) Fencepost errors are irritating.

honey the codewitch 22-Feb-23 6:31am View

See, the problem is I get ambitious and then things like the OP happen. :(

honey the codewitch 22-Feb-23 6:24am View

m_controls holds a vector of controls sorted by z-order, back to front. The screen background is redrawn with bmp.fill(...) in the code.

honey the codewitch 22-Feb-23 6:23am View

The problem with asserts on this platform is you don't get a reasonable debugger (JTAG makes you want to get out and push) - asserts will create a boot loop which actually makes things harder to debug. Better to dump to the serial port, or failing that, while(1); the damned thing after dumping a message to serial. I'm dumping to the serial for this, but I'm not 100% sure what to expect so I have nothing to feed an assert.

honey the codewitch 22-Feb-23 6:21am View

I hate this. You don't know how many times I've run into questions I can barely ask, much less find someone who can help with the answer.

honey the codewitch 22-Feb-23 6:20am View

It would be twice as much code to abstract this time critical function, and designing an interface between high and low level would be almost as difficult as creating this.

honey the codewitch 22-Feb-23 6:20am View

The algorithm already does that. It's the second part of the routine.

honey the codewitch 22-Feb-23 6:19am View

Each control has an invalidate() function that marks it for add in the dirty rectangle list built at the start of the routine. I don't offset in the render routine of each control. I offset all draws in an intermediary canvas of type control_surface_type which sits over the bitmap and offsets all drawing operations. All drawing operations to a bitmap are automatically clipped to the bitmap. None of this is my issue. It's just positioning.

honey the codewitch 22-Feb-23 1:20am View

Yeah. I understand. I just ate a BLT and intend to nap. Maybe a solution will present itself that way.

honey the codewitch 24-Dec-22 21:25pm View

Hey, whatever works

honey the codewitch 24-Dec-22 11:18am View

No problem at all. Glad to help. I love little C++ brain teasers =)

honey the codewitch 29-Nov-22 5:00am View

Thanks! I've kind of answered my own question since I asked it - funny how that seems to work so often, but now I'm interested in what others have to say. =)

honey the codewitch 30-Aug-22 12:02pm View

code generation is not a library, nor can a minimal IoT device function with a UI library on battery power.

honey the codewitch 30-Aug-22 9:19am View

Yeah. I've been coding since 1986, professionally since 1996, so I hear you.

honey the codewitch 29-Aug-22 11:16am View

I can't have it all, but I can endeavor to have the important stuff. I have a pretty solid instinct for when the design crystalizes and I can move forward with development, and I just haven't banged on this design enough yet. I'll figure something out that gives me everything I actually need, and I have the time to drag this out if it means getting it right.

honey the codewitch 29-Aug-22 11:13am View

You're not wrong. In fact, I was thinking about a fixed scheme like you mention, and on reflection I realized that I would know how many widgets there were in advance. But runtime modification of said widgets is where I run into trouble. For example, as soon as I make it so you can change the text on a label at runtime, the complexity gets much higher. The reason being is I have to switch from laying things out on a screen by screen basis, to an "window/control" by "window/control" basis, and that means computing dirty rects and stuff, or using a painters algorithm on a godawful slow device.

At least that's where I'm at with my thinking now. I still haven't decided, because I'd prefer (A) if there was a good way to skin that cat, but (B) seems to be what I am leaning on in absence of that better way, so maybe I just need to spend some more time thinking about (A)

Edit: Hahaha I even swore in my comment. Now you know where my head is at.

honey the codewitch 26-Aug-22 10:52am View

That's good advice for desktop apps, but "general purpose library" and IoT are generally at odds with each other. Every abstraction costs, and when you're dealing with 80kB of RAM and < 100MHz sometimes it matters. Also, projects can't be very large on most devices.

I'm wondering if there isn't a way I can separate the generated code from application code that uses it, sort of like Microsoft's code-behind paradigm. That might give me the best of both worlds.

honey the codewitch 16-Aug-22 11:00am View

That can't happen because of the way these streams work. buffer_size will always return buffer_capacity except for the final block, after which it will always return zero. Therefore, it never hangs. And yes, a while loop would be better, but I learned something in martial arts that serves me here - first get it done, then do it quickly. Anyway, I found the bug, and it wasn't in this code. Of course. Thanks anyway.

honey the codewitch 31-Jul-22 16:01pm View

The buffer logic implementation is almost all in read(). That's what gets me. That's what does all of the heavy lifting, and as far as the underlying stream, I've tested it. It works flawlessly when I feed it directly to the downstream consumer (a jpg loader) but when I add the buffered stream it croaks. Interestingly, it works for loading TrueType font streams, ***but.. TrueType font streams seek all the time, which resets/clears the buffer, so it's effectively not using the buffer. Just some background.

honey the codewitch 31-Jul-22 12:33pm View

I'm pretty sure this is correct (I'm using gcc) but I can always break it out into a full if.

honey the codewitch 31-Jul-22 12:32pm View

It actually is recursive, but it recurses only in the case where the requested amount is greater than the buffer capacity. m_stream is the underlying stream.

honey the codewitch 24-Jul-22 10:29am View

Yeah I solved it this way earlier thanks to Peter. I just wasn't able to come at the problem the right way. Funny how something so simple can get so complicated.

honey the codewitch 24-Jul-22 10:28am View

Whoops. my reply was meant for Griff

honey the codewitch 23-Jul-22 19:21pm View

You're overcomplicating it. The hardware is doing the timing. I am sampling at 44100Hz. With that in mind I am computing a sin wave based on a rolling phase that is a derivative of that sample rate and the specified frequency. const float delta = TWO_PI*frequency/sample_rate;
I then use that delta to increment my phase between 0 and TWO_PI by delta each time, generating a sort of sawtooth that I apply f=sin(phase) to to get my wav. The *only* reason I can't generate that as a triangle given what I know how to do is because the phase is a saw rather than cycling up and down and up and down. But the sin *does* cycle up and down, so I was hoping I could use that. My square wave is just f = phase>PI; So I don't know. I want to get a triangle by deriving one of those.

honey the codewitch 23-Jul-22 17:58pm View

Oh okay. I *think* I get it. Time to go get dangerous.

honey the codewitch 23-Jul-22 17:47pm View

but if i add up the square waves they will compute a tri wav at a much lower frequency and much higher amplitude. The amplitude I can scale, but the frequency issue is problematic. I was hoping I could just "straighten out the curves" in the sin wave and work from that. Barring that, I'd need to bisect the square wave. Computing tangents I guess would work? But I'm not great at math. I was in my 20s before I learned that trig was based around triangles. *hides*

honey the codewitch 6-May-22 13:33pm View

When you're compiling for little devices -Os is in effect on GCC.

honey the codewitch 6-May-22 8:51am View

*facepalm* I forgot that shift left and shift right are endian dependent. stupid stupid stupid. i know better.

honey the codewitch 6-May-22 8:42am View

that's weird to me, because Most significant would refer to the value that affects the high numbers, and Least significant would be the value that affects the low numbers no? That would be different for machines of different endian types, unless I'm missing something. It's frightfully early and I'm still catching up on coffee.

honey the codewitch 6-May-22 7:53am View

It shoves the bits high if it's little endian - as seen by the shift, otherwise it shifts them low.

If it shifts them high, is that MSB or LSB? That's what I'm trying to figure out. Right now I call it MSB

honey the codewitch 6-May-22 7:18am View

I'm confused. "Take, for instance your msb function (I didn't scrutinize the other ones), it takes value and stores it in the second byte of value16 on a little endian machine."

Masking off the 8th bit is part of the midi spec.

That's what it's supposed to do. My only question is is that LSB or MSB?

honey the codewitch 6-May-22 0:04am View

That's not my question though. I already do those checks at compile time. Those ifs are not evaluated at run time.

What I want to know is if my msb() and lsb() accessors are named correctly from big endian perspective, or if I have them backward

honey the codewitch 2-May-22 8:38am View

As long as it's clear, I'm cool with it. I do weird things like that too. :)

honey the codewitch 2-May-22 8:01am View

Nope. It's shorter. I'm increasing the top y value, so it starts further down the screen, and since y2 is unchanged that makes the rectangle less tall. I swear to you it works, I'm looking at it running right now. Remember that all the rectangles are effectively anchored to the center on their bottom edge. That might be an area where I should clarify what I'm doing.

honey the codewitch 2-May-22 7:45am View

Thank you for your feedback! GFX is targeted more toward intermediate to advanced developers and those that actually do IoT for work like I do so some familiarity with "real"/modern C++ is assumed. Basically once you've graduated from Arduino IDE and want to do something more powerful - like utilize the capabilities of that fancy ESP32 you just bought, then GFX is the way to go. LVGL is comparable featurewise, though I don't provide widgets, and I *do* provide full TrueType support, even under Arduino. LVGL can support truetype in limited instances, and it uses gobs of RAM and the firmware must be compiled on a linux machine. Also only targets the ESP32 WROVER. Mine targets whatever GFX does. =)

honey the codewitch 2-May-22 7:42am View

First, GFX doesn't work with the Arduino IDE because it won't compile with C++11. I use the platformIO IDE with it, which runs inside VS Code so you get all that autocomplete goodness, and just far more control over your project builds than you can get with the Arduino IDE.

Spoiler on the code:

Yeah, the clock basically forces the rectangle you pass it to be square by looking at the shortest edge and treating that as both the width and the height. That's what w is.

Then it *is* centered. horizontally. See, all hands are drawn effectively at the 12:00 position (before rotation), and then rotated based on the corresponding time figure.

Notice the center of the viewport is set at the beginning. That means the rotation will be around that point. =)

honey the codewitch 2-May-22 6:35am View

What I'm trying to do is create examples for people that don't read the documentation.

A lot of times people that use arduino are hobbyists and don't RTFM.

Assuming autocomplete, and the above, can figure out what it's doing from the context?

Not necessarily precisely - just imagine you're looking at a code sample to figure out how to do something. How hard is this one to figure out enough to fiddle with it, if that makes sense?

honey the codewitch 2-May-22 5:44am View

hahaha I had changed the order above after-the-fact for readability and it's early. I fixed it.

honey the codewitch 25-Apr-22 6:54am View

That's what I thought too so I recloned the git in a separate project. Same deal. I also am manually checking the compiler error line # with the line # shown by VS Code and it matches.

I'm starting to think there's a bug in the GCC compiler? I need to rename my variable again to see if I wasn't hallucinating.

honey the codewitch 25-Apr-22 6:53am View

it's just a temporary holder for an error return value. Typically for things like this I use one letter names. It actually makes it clearer because it's easier to scan visually what is "throwaway" and what is part of the actual algorithm.

There are two reasons I know I'm not looking at the wrong code. For one VS Code provides line #s and click to jump to the location of compiler error messages. I've also manually verified.

honey the codewitch 19-Apr-22 18:05pm View

I wouldn't be surprised if it's off. I was actually trying a number of things. CPallini caught a show stopper for me that explains why i got nothing.

honey the codewitch 19-Apr-22 18:04pm View

I shouldn't said the thing about the top left. I want the colors to start at the start color and radiate away from the center point where the start color is, toward the extents of the rectangle, where at least somewhere, end_color should be represented. This is actually easier to do spherically, but I want to do it so it is along an imaginary line instead of radiating away from a point, if that makes sense.

honey the codewitch 19-Apr-22 18:01pm View

If rect.y1>rect.y2 xd will be -1, oooooh thank you. I just saw it.

honey the codewitch 19-Apr-22 15:49pm View

I learned a long time ago to first make it work, then make it work fast. It serves me well, except if there are design issues tied up with performance. The blast radius of any design changes are limited pretty much to this routine, so if i have to rewrite it for performance I will. I wrote this to try things out, and I didn't want it too laden with premature optimizations.

Edit: Looking at your solution, it's too simple for my application, as it only does a gradient in one direction, not at arbitrary angles with arbitrary starting points.

Pixel blending by channel is something my library already does when you call blend.

honey the codewitch 1-Apr-22 8:40am View

Have you considered studying? Imagine you "earned" your degree by way of other people and landed a programming job. Are those other people to do your work for you as well? Maybe if you're going to school, take advantage of it.

honey the codewitch 26-Mar-22 21:59pm View

Yes that would be correct. As far as your solution, I'll have to think about it. It does give me an idea. There's a chance it will complicate another part of the code and right now I'm too foggy to determine whether it will without diving into it.

honey the codewitch 17-Mar-22 3:29am View

That's one of the reasons I haven't really bothered doing much fixed integer arithmetic. There is one area in that routine where it does, but it makes sense to because it's just blending RGB888 color channel values.

honey the codewitch 16-Mar-22 18:12pm View

Yeah unfortunately optimizations like this won't take my minute down to seconds, which is what I need. I need orders of magnitude of performance improvement.

honey the codewitch 16-Mar-22 14:19pm View

Starting at line 573 here
https://github.com/codewitch-honey-crisis/gfx_demo/blob/master/lib/gfx/include/gfx_palette.hpp

the tables are here: https://github.com/codewitch-honey-crisis/gfx_demo/blob/master/lib/gfx/src/gfx_palette.cpp

I'm using "mixing_plan_fast" although I have two.

Sorry for the length and interdependencies. This code is not really user friendly.

honey the codewitch 11-Mar-22 13:04pm View

That's basically where I'm at right now. I'm going to keep this open to see what other ideas there are out there before I put a lid on it. Just in case there's some cleverness yet to be mined kicking around in someone's head.

honey the codewitch 11-Mar-22 4:45am View

I totally didn't even see the link when i first read the question. must have been having a day.

honey the codewitch 10-Mar-22 13:23pm View

I second PIEBALDconsult here. Where is your stream coming from? If it's coming from a network source Peek() will fail. You really shouldn't Peek() if you want your code to be robust. Use Read() and just keep a copy of the current character around in a variable instead of using peek.

honey the codewitch 8-Mar-22 17:17pm View

Earlier there was an error in my question. I tried two different ways, in the question, and have settled on this
(R-R2)²+(G-G2)²+(B-B2)²

See the edit to the question. The code was right, but my description of what i was doing was wrong in the question before I edited it. Sorry.

Edit: It turns out it was an unrelated bug. My color matching code worked fine.

honey the codewitch 8-Mar-22 16:27pm View

I'm working on figuring that out. I had to take a cat nap =)

Edit: I've edited my question significantly to update where I'm at.

Edit 2: There was nothing wrong with my code to begin with. It was an unrelated bug.

honey the codewitch 8-Mar-22 15:19pm View

And I did check it. Turns out the formula in my question is wrong, but in the code it isn't. I use a variant of the Euclidian method sans the final sqrt() since I don't need it (i'm only comparing relative distances so if i sqrt none of them it's the same as sqrting all of them). I've since updated the question with the proper formula. thanks for catching that. unfortunately it doesn't fix my code. =/

honey the codewitch 8-Mar-22 14:47pm View

the first method is a variant of the Euclidian distance method. The final sqrt is removed as it's not necessary in this case.

... and as I was typing this I realized I was doing it wrong above compared to the wiki.

so thank you for probably solving my issue, albeit just by making me check my work.

honey the codewitch 6-Mar-22 19:31pm View

The problem there is I need to friend the other way around, and I don't have all the template arguments.

honey the codewitch 5-Mar-22 12:21pm View

This is the Arduino framework. setup() is the entry point, and loop() is called repeatedly by the framework for the life of the app once setup exits. There is no exiting "main" - main is implemented as main() { setup(); while(true) loop(); } basically (simplified)

honey the codewitch 8-Jan-22 1:03am View

Happy new year to you as well.

honey the codewitch 8-Jan-22 0:38am View

Yeah, I'm not trying to execute the regular expressions. I'm trying to simplify and compact them programmatically.

honey the codewitch 7-Jan-22 8:25am View

I should have clarified. This is DFA regex. No backtracking constructs. I can already convert to a regex string from a state machine. this is only about simplifying the result.

honey the codewitch 7-Jan-22 8:25am View

Deleted

I should have clarified. This is DFA regex. No backtracking constructs. I can already convert to a regex string from a state machine. this is only about simplifying the result.

honey the codewitch 25-Dec-21 1:28am View

You're not hearing me. Even if I provided you an example, you would not understand it, and so it wouldn't benefit you. If you were to use it, you could not maintain that code, and I'm not going to maintain it for you.

Therefore I will not give you that. It would be like giving someone my loaded gun.

This is not something you should be trying before you even understand even the basics of how to code C++. Learn how to code some C++ first.

honey the codewitch 24-Dec-21 19:54pm View

It's because you're using cdecl as your calling convention. And how is C# supposed to know the size of your array? You'll have an easier type using stdcall as your calling convention, and yielding an out pointer to the array and an out pointer to the size of the array for your function.

your C++ code is invalid anyway. Anything you declared in the function goes away once the function is complete, meaning you're returning a bad pointer.

Also using unsafe code in C# is inadvisable, because unless something changed, it tends to require full trust, so it won't run unless it's installed in a particular way on the target machine.

I'm going to be very frank with you. You are swimming way out of your depth right now. Learn C#. Learn C++. Then learn how to use them together. Right now you don't know any of them, except maybe some C# from the looks of things.

honey the codewitch 14-Dec-21 13:52pm View

It's baking my noodle. I'll hit you up if I have questions. I'm not accepting just because I want more ideas but I'm +5ing your solution. :) Thanks.

honey the codewitch 8-Dec-21 13:59pm View

I'll look into that, thanks.

honey the codewitch 7-Dec-21 11:10am View

Totally. This code flips physical digital pins on a chip high and low, which it needs to do fast and precisely timed, because it's a driver for an 8-bit parallel bus attached to a (usually color) LCD or TFT display. The frame rates I can achieve with graphics draws are directly tied to this code.

honey the codewitch 7-Dec-21 9:10am View

Hmmm.. It seems a bit sloppy to me because I don't like the idea of introducing arbitrary template arguments just to templatize a function. The code smell, you understand. If I come along in 6 months and every function is templatized with dummy arguments, I don't what I'd do. I think I'll stick with the __attribute((always_inline)) route suggested before, because I'm targeting GCC specifically.

honey the codewitch 6-Dec-21 18:13pm View

That could work, so I'm accepting it, since I wasn't familiar with the attribute prior. I'll give it a shot. I have to optimize for size. I'm not even sure how my toolchain would react otherwise, I just can't afford the extra space, particularly on STM32 boards. I can't use clang because my toolchain relies on GCC and I don't have any control over that. Actually all my IoT toolchains require GCC at the moment.

I do like to compile with both clang and GCC (when i'm not doing IoT-only) but GCC is my primary target.

honey the codewitch 20-Nov-21 5:33am View

I'd have suggested they continue to ask the question until their code magically works.

honey the codewitch 15-Nov-21 13:31pm View

Why did you take a job you're not qualified to do?

honey the codewitch 10-Nov-21 11:54am View

Can you post several example strings, with example matches you'd like to see?

List several variations of this (including one with no matches)

And then below each one, list on a separate line each substring you'd like matched.

AB-0001234/00005-Testing My Regex hereAB-0004567/00007-Second Testing of my regex
--- matches ----
AB-0001234/00005
AB-0004567/00007
Second
Testing

(assuming you also wanted to match Second and Testing - i know you don't, i'm just trying to give you an example)

Anyway, make at least 4 of those, including one with no matches, and I bet you'll have a better time getting an answer.

honey the codewitch 9-Nov-21 22:35pm View

Thank you! That clears it up for me perfectly.

honey the codewitch 5-Nov-21 5:02am View

That was my original plan for Python. I just wasn't sure if it was valid. Actually right now I think my multiline conditionals are indented once rather than twice but it's an easy fix in my generation template. the big issue is knowing what's valid, and secondarily what's pretty. I'm on the fence about double indenting multiline conditionals hanging off of ifs. It's prettier to corner case it, but I don't like it because it's not consistent with the general rule of indenting I follow through the rest of my code. It just rubs me wrong.

honey the codewitch 5-Nov-21 4:34am View

Not in that case, but there are other cases where it's needed. It makes code generation a bear. Fortunately in my current case the blast radius of this problem is limited to if () statements.

honey the codewitch 5-Nov-21 3:08am View

Thanks, I totally blew by the link in the reply. It's the hour that it is. I'm not altogether with it right now. =)

honey the codewitch 5-Nov-21 1:29am View

Are you sure it's not indented at the next level? My google fu isn't helpful right now, and I don't want to stop what I'm doing to start an install of that mess. Oh well. I'll figure it out one way or another. Thanks.

honey the codewitch 11-Jul-21 11:55am View

I can't target C++20 unfortunately. I'm limited by the minigw gcc version used by platformIO under windows (the least up to date build platform) and that means C++14

honey the codewitch 11-Jul-21 11:54am View

> Can't you just say that instantiations must include various has_blah functions

In some cases, maybe if I had designed it that way. However, because things are typed to a pixel definition and other things, you can't use bool functions for every case. You have to use usings in some cases. Since I already have a using for this case, I find it inelegant to require the draw target implementer to create function to say essentially the same thing.

honey the codewitch 11-Jul-21 8:28am View

I have an idea though. I'm wondering if I can't go template<typename T,size_t NativeBitDepth=T::native_pixel_type::bit_depth> struct test_for_native_pixel_type { ... } or something like that. I don't know what the compiler will do there if native_pixel_type isn't declared. My assumption is that it would error but perhaps it wouldn't. I need to check. It would only be zero if native_pixel_type was not present and the compiler had to choose a default for size_t. It *could* work, i just don't know if it will.

honey the codewitch 11-Jul-21 8:21am View

I have "draw destination" classes which are loosely defined (bound via template) and have an interface/contract that includes usings. One of the members might be something called native_pixel_type and if it's present it tells me that "virtualization" is being used which means it's dithering. If it's dithering I don't want to use anti-aliasing to smooth the edges of fonts because there's not a true grayscale to work with, just black and white.

honey the codewitch 25-May-21 11:50am View

Whoops, I forgot to increment my fill pointer. Not sure what i was thinking. Still, it doesn't solve my issue but thanks for spotting that.

Edit: Never mind. retr() will fill "fill" each time with new data. The y is just there to make sure it gets called the right number of times.

honey the codewitch 25-May-21 10:57am View

The reason it's defined that way is it's 16-bit color values, but send and retr take a byte array. It's fine - casting does nothing to the data. It only changes how the data is perceived. It is a "pure" operation with no side effects (unless it is overloaded, and in this case it isn't). say you had a 16 bit word of 0xC1A2. If you cast that to bytes you'd have a byte C1 followed by A2 (actually reversed on little endian systems but you get the idea). send and retr deal in BLOBs which are byte arrays, effectively. But it's easier to deal with pixels in their native format, hence the discrepancy there. a 16 bit pixel is R:5, G:6, B:5 so you can't split it cleanly.

As to your second question, each color is 2 bytes. The pixels are 320 across. The compiler will "fold" the value to 640 with no runtime overhead, but defining it this way makes it clearer that it's 320 pixels of 2 bytes each. It's just a way to express intent with code.

honey the codewitch 24-May-21 8:59am View

If he does, he'll regret it. I didn't post the answer for him. I posted it because it was a good question. I half assumed he'd be kicked for the autism remark.

honey the codewitch 10-Apr-21 7:37am View

Thanks. I actually tried that after I posted the question, but I had the syntax wrong. I ended up writing a helper function to disambiguate. It's for the best anyway that I did it that way, in the end. Still, this is the proper solution.

honey the codewitch 2-Apr-21 3:19am View

Prototype it in C and then simply spit the asm out using the compiler.

There's hardly any reason to use an assembler directly anymore, and many good reasons not to, but if you *must* use an assembler, use C or C++ to prototype your assembly code where you can.

The above can easily be written in C, so write it in C, and then use your favorite C compiler to get the assembly from it, and then look at that.

You haven't specified which assembly you want. Are you coding for an ATmega2560, a Tensilica, an ARM, x86?

Don't bother. Just use C and get asm that way. If you want to tune it from there you can.

honey the codewitch 17-Mar-21 12:17pm View

Okay, it sounds like what you're really trying to do is access something like a control from outside the form it's declared on. For example, you have class a that needs to access an instance of Form.cs. Where is Form created? Because first you'll need that. If it's your main form, it's created in Program.cs in the Main() routine as part of Application.Run().. it calles "new Form()" - you will need the value it returned to access any fields in it. It is simply not otherwise directly possible since that form could be anywhere in memory. Or there might even be more than one of them, or none of them. Without some sort of access to the "pointer" for the instantiated Form you are out of luck. Form is just a class. It's nothing until it's created with "new"

honey the codewitch 3-Mar-21 12:53pm View

src/vfs.hpp:152:31: error: invalid conversion from 'int (*)(int, stat*)' to 'const void*' [-fpermissive]
memcpy(&test1,vfs.fstat,sizeof(void*));

I can't set the permissive flag because gcc is run as part of a large, complicated toolchain and it doesn't like to be touched.

honey the codewitch 3-Mar-21 12:38pm View

Yeah i caught that a minute ago. it amazes me i let that go. I've updated the code. it still doesn't compile.

honey the codewitch 22-Dec-20 10:03am View

I know i know. it's early and i'm sleepy and typos and poorly worded thoughts flow freely from me right now.

Edit: (per yours), Yeah I may need to break out paint!

honey the codewitch 13-Dec-20 11:29am View

Normally I don't find much occasion to mix them, but I wanted a more compilery, less preprocessory way of defining a maximum capacity for a fixed length array.

I'm not aware of a better solution to the problem than what I've done above, in this particular instance. The code difficulty is limited because the class isn't that complicated. This was one wrinkle, and it was a compiler SNAFU. A good reason not to use it sure *if a better replacement can be had* but I'm not aware of one for this case.

honey the codewitch 13-Dec-20 10:37am View

Totally. They can be frustrating at times, but Accelerated C++ changed my life and I do a lot of generic programming now. Less so on Arduino stuff but still.

honey the codewitch 13-Dec-20 10:32am View

nah it should be able to forward precisely *because* it's virtual. there's a vtbl in play so it just needs to find the function.

Turns out the compiler was looking for a global function called initCapture().

Since it's part of a template there's a 2 phase resolution process in C++ and it wasn't picking it up as part of the class.

So if you do this->initCapture() it all works fine. It gives the compiler what it needs to look in the right place (the base class) for the function.

honey the codewitch 11-Dec-20 19:59pm View

Wonderful! I'm so glad when my stuff can help someone. We all start somewhere. Good luck with your project!

honey the codewitch 11-Dec-20 9:45am View

unique_ptr takes the type and decorates it so you're creating a pointer to a pointer. Solution 1 is the correct answer.

honey the codewitch 1-Dec-20 7:07am View

You've said nothing about the problem. You've just mentioned that a problem exists. How is anyone supposed to help you when they don't know the nature of the problem you are running into? "Licensing issues" is not a description of the problem, and most people may not have heard of "black duck scan"

honey the codewitch 30-Nov-20 20:36pm View

You can't use windows forms in core 3.0. Either way you're going to need to create a window from the looks of it. It might not need to be visible. That, and the code I offered is what I have. If you want to know about TWAIN you'll just have to study it as I don't know anything about it.

honey the codewitch 29-Nov-20 12:48pm View

Hey thanks. This is awesome! You don't mind if I use a derivative of this code in a commercial gadget do you?

honey the codewitch 29-Nov-20 11:47am View

Thank you!

honey the codewitch 29-Nov-20 9:23am View

I can't do it that way because readNextPoint() reads from flash which is very slow. I need to do it one sample at a time to keep things running smoothly, if I can at all.

Therefore if my solution gets me the same as your solution I prefer my way. I actually endeavored to avoid doing it the way you do.

I should add that the number of points is compile-time configurable depending on the defined resolutions so sometimes that loop might be much larger. =)

Comments by honey the codewitch (Top 200 by date)