|
You might already be familiar with callbacks and not realize it. If you've ever used qsort() or bsearch(), or another library function that takes a function pointer as an argument to help it do its work, you have already used callbacks!
For example the signature for qsort is
void qsort(void *base, size_t nmemb, size_t size, int (*compar)(const void *, const void *)); and might be used like this
struct foo {
char key[10];
int id;
};
int cmpfoo_bykey(const void *vp1, const void *vp2)
{
const struct foo *foo1 = vp1;
const struct foo *foo2 = vp2;
return strcmp(foo1->key, foo2->key);
}
int cmpfoo_byid(const void *vp1, const void *vp2)
{
const struct foo *foo1 = vp1;
const struct foo *foo2 = vp2;
return foo1->id - foo2->id;
}
int main()
{
struct foo my_data[100];
qsort(my_data, 100, sizeof(struct foo), cmpfoo_bykey);
qsort(my_data, 100, sizeof(struct foo), cmpfoo_byid);
return 0;
}
In the above example qsort() knows how to sort an array of items in a generic way, but not how to compare items. It gets around this limitation by using a callback function, that the programmer provides, that does know how to compare items in the array. So, whenever qsort() needs to know how two items in the array compare to each other, it calls back to the code provided by the programmer to determine how the two items compare to each other.
Keep Calm and Carry On
|
|
|
|
|
I'm programming in C in an embedded environment and I have a bit-bang software UART I would like to be able to hook up to a myPrintf function, that supports the standard printf syntax, e.g.
myPrintf("%s %3.2f\n", "Pi:", 3.14159);I would like to achieve this without having to use a big (=worst case length of "sentence" to transmit in a single myPrintf call) temporary buffer (otherwise, I could just go via sprintf). What I want to achieve is that my characters are sent over my UART "on the fly", so in the example above "Pi: " would be sent out over my UART before the float number has been converted to ASCII characters. Does anybody know how I can achieve this in the easiest way possible? Is the source code for printf available somewhere for public use so I can modify it to fit my needs?
|
|
|
|
|
Quote: Does anybody know how I can achieve this in the easiest way possible?
Not sure if easiest but one common way of doing it is to have printf call a putc-type function to send out each succesive character. In your case, the "putc" function would be the UART output function.
Quote: Is the source code for printf available somewhere for public use so I can modify it to fit my needs?
Plenty of those. One I've used before in an embedded environment is for Rabbit microcontrollers. You can find the relevant code here: DCRabbit_10/STDIO.LIB at master · digidotcom/DCRabbit_10 · GitHub[^]
Mircea
|
|
|
|
|
Mircea Neacsu wrote: Not sure if easiest but one common way of doing it is to have printf call a putc-type function to send out each succesive character. In your case, the "putc" function would be the UART output function.
That sounds very convenient. I am working with STM32F-microcontrollers, would you happen to know if it's easy to replace the putc-function for their printf-function?Mircea Neacsu wrote: Plenty of those. One I've used before in an embedded environment is for Rabbit microcontrollers. You can find the relevant code here: DCRabbit_10/STDIO.LIB at master · digidotcom/DCRabbit_10 · GitHub[^]
Awesome!
|
|
|
|
|
Quote: I am working with STM32F-microcontrollers, would you happen to know if it's easy to replace the putc-function for their printf-function?
No idea! you said you have the bit banging UART code; you just have to use it to send out one character.
Keep in mind also what @CPallini suggested: if you don't need a very general printf function you can have with a much simpler implementation. Look at this one for an example: sdcc/printf_small.c at master · darconeous/sdcc · GitHub[^]
Also, along the same lines, you scan the format specifier string and every time you encounter a '%' character you can use sprintf to print it in a small buffer.
Mircea
|
|
|
|
|
Probably existing printf code is too general for your system, so, it is probably better to make yourself an over-simplified version of it, tailored on your needs. Anyway, you know, in order to use just a small buffer you need to append characters to the buffer while there is still available space and wait patiently the transmission of characters in order to use the buffer again. This means you can't actually use a function call similar to the printf one.
"In testa che avete, Signor di Ceprano?"
-- Rigoletto
|
|
|
|
|
CPallini wrote: in order to use just a small buffer you need to append characters to the buffer while there is still available space and wait patiently the transmission of characters in order to use the buffer again. This means you can't actually use a function call similar to the printf one.
It's a completely synchronuous bit-bang UART where I'm counting clock cycles. There is no DMA, timer interrupts, 1 byte FIFO, threading or anything fancy.
|
|
|
|
|
So you have to it in place. Forgot the printf signature and write a function to output a string, a character at time. Then write a function to output properly a floating point number (that would be more difficult, but it depends on your requirements) and combine the two function in order to obtain the expected result.
"In testa che avete, Signor di Ceprano?"
-- Rigoletto
|
|
|
|
|
I made an app that used to works just fine to change gamma, and it still does, but with one condition, to have only one monitor connected. But the problem now is that I have 2 monitors, and the app doesn't work anymore, only if I disconnect one of them.
Can anyone tell me what I need to change in order to make it work with multiple display setups? And if there is a way to change gamma on only one of them then that is a good bonus and will help with another idea that I have.
This is what I'm using right now:
WORD ramp[256 * 3];
for (int i = 0; i < 256; i++) {
ramp[i] = ramp[i + 256] = ramp[i + 512] =
(WORD)min(65535, max(0, pow((i + 1) / 256.0, ValueForLight) * 65535 + 0.5));
}
SetDeviceGammaRamp(::GetDC(NULL), ramp);
|
|
|
|
|
|
I don't get it. I don't see anything there that may interfere with the app and the way I have my PC.
|
|
|
|
|
The first paragraph clearly states that you should not use it, and it may even fail without giving any indication, and indeed returning TRUE suggesting that it did not fail.
|
|
|
|
|
The following statement is a personal opinion and is not addressed to you specifically! So take it as it is!
These types of answers are actually more harmful then they are helpful. Not only that they don't actually help the person reach a solution but they may affect some, and actually will affect certain people by at least demoralizing them. We are not in school to say "I can't give you a straight answer because you are on a test" or by giving them a riddle which may lead that person on a wrong path that is even farther from the solution, we are here to learn from each other. So it is best to actually not to say anything then giving an answer that isn't helpful or by making fun of them for not knowing. Not everyone is good at searching on google for the answer, maybe they don't know exactly what to search for. Why is it so hard to tell someone, "Here you do it this way, use that and like that"?! This as a side note, because I saw in general, on other platforms as well, many answers that were so bad and actually harmful. I have a friend that tried to get into programing but quit it because he posted like only 2 questions on another popular website, and because the questions where really beginner level many started criticizing him for what he asked, like "How can you not know that?!", well he was just starting so how could he have know.
As for what you said that, in that paragraph it says not to use it, yeah it is true, but it also says the following "Use of this API is subject to major limitations", so it's not like not to use it because it is bad, but because other reasons.
As for the other part, it is actually a good thing in that situation that it doesn't uses what you give it, and it actually is telling you why in documentation, because it may lead to an unusable screen.
|
|
|
|
|
Richard posts a link to the documentation where it clearly says "We strongly recommend that you don't use this API" and then you complain that it isn't the kind of answer you wanted ?????
The documentation even goes so far as to tell you how to do what you want using a supported alternative method!!
DID YOU EVEN READ IT!?!?
|
|
|
|
|
First of, I wasn't talking about him giving a link to the documentation. That is a good replay when you don't know a solution, but you still want to try to help. I did seen it before posting my original questions, but it doesn't say anything about the problem I encountered. I even mentioned that it was a general opinion, and at the end of the statement I added something as a respond to what he said.
So back to the matter, yes they do say not to use it, but they also give you reasons why not to, which none of them are bad for the use case the app I made is meant for. So by taking in consideration their warnings in the documentation, it was still a good option to use.
Quote: The documentation even goes so far as to tell you how to do what you want using a supported alternative method!!
Yes, it does talk about some alternatives, but I don't need color calibration, or blue light filtering, or color adaptation.
Quote: DID YOU EVEN READ IT!?!?
So based on what I just said, I can ask you the same, did you read everything I said?! But as I said in that statement, this type of behavior isn't helpful so why not try to avoid it, and actually try other means of expressing to get your point of view understood, as this is something optional to do, no one is forcing you to answer every question the users of this website are posting?!
I even saw this article in which they talk about evolution of methods to do that, but there are a few problems with other methods they present, first and most important this:
Quote: The Direct3D 9 and DXGI APIs are “write only”. You can't read the value of the hardware, modify it, and then set it. You can only set the ramp.
The problem with this is that my app is meant for general gamma change, which means when the user closes the app, he has an option to restore to the original gamma value, which with those 2 you can't.
And also this problem:
Quote: Furthermore, you can only set gamma when the app is full screen.
I need it to work in general not based on an app that is running in fullscreen mode.
|
|
|
|
|
I'm sorry, did you want someone to click on the Windows Color System[^] link for you?
That's what you're going to have to use.
|
|
|
|
|
I'm sorry but how is that theory goin to help someone who doesn't know what code to use? OK, I'm learning the theory behind it, but that isn't goin to help me write the code. I never said that I have a good knowledge when it comes to C/C++. Actually it is kinda bad, that is why this function is goin to be called as a native one.
Just out of curiosity, have you ever thought that maybe there are some that understand better by seeing it in practice then reading about the theory behind certain things? And surely I'm not the only one, after all there is this saying "Practice makes perfect".
|
|
|
|
|
You apparently didn't look at the rest of the page, especially the navigation down the left side, and completely missed the link to "Reference", which gives you all of the functions involved with WCS and the documentation on each.
If you're looking for someone to write the code for you, or to copy, I can't help you there. I have no need of color correction at all and don't have the time to the someone's research for them.
|
|
|
|
|
Valentinor wrote: will affect certain people by at least demoralizing them.
So are you saying it is demoralizing to be given a link to the documentation?
Valentinor wrote: Not everyone is good at searching on google for the answer
I just Googled for the name of the API function that is in your question; why do you think that is difficult?
Valentinor wrote: As for what you said that, in that paragraph it says not to use it, yeah it is true, but it also says the following "Use of this API is subject to major limitations", so it's not like not to use it because it is bad, but because other reasons.
As for the other part, it is actually a good thing in that situation that it doesn't uses what you give it, and it actually is telling you why in documentation, because it may lead to an unusable screen.
So you have the complete answer, why complain?
|
|
|
|
|
Quote: So are you saying it is demoralizing to be given a link to the documentation?
I wasn't talking about that answer. Giving a link to documentation is a good start when you don't really know the answer, in hopes that the person who asked the question didn't checked it before.
Quote: I just Googled for the name of the API function that is in your question; why do you think that is difficult?
So with that I wasn't talking about me, it was in general, everything in that statement beside the part at the end, was something I observed over the time.
Quote: So you have the complete answer, why complain?
I didn't said the problem is solved. In the above message in responds to what Dave Kreskowiak said, I went into more details about the problem, as no one even tried to ask more question to understand the problem better, in case maybe it wasn't explained properly, which the founder of this website actually encourages to do so.
Quote: If a question is poorly phrased then either ask for clarification...
I'm glad that you actually read the whole thing, even tho there were some misunderstandings about the message it was meant to deliver.
|
|
|
|
|
(C++ and Visual Studio 2019)
We have a 3rd party library that we want to use (we build from source and it generates many, many static libs).
I want to create an abstraction layer on top of this 3rd party library, as a static library.
And I want my client application to use my own abstraction layer so it does not have to directly include the 3rd party headers.
Can I "link" the 3rd party libraries (libs) in my own library so that I can just link we one library in my client application ?
Or I will need to link my application with the 3rd party libraries as well as my own library ?
I'd rather be phishing!
|
|
|
|
|
If you create a static library that links to other statics then the result should be a single static library. A simple test should confirm it for you.
|
|
|
|
|
Crikey
I did not even look at that (bad case of mondaying)
It seems to be working.
Thanks Richard.
I'd rather be phishing!
|
|
|
|
|
Yes, it's the sort of question that gets you overthinking in your head. 
|
|
|
|
|
You can create a "composite" that includes your modules and all modules from 3rd party libraries.
About including the 3rd party header files, it depends how good of a job you do at hiding the internals from your users.
Thing to keep in mind: remove any #pragma directives that reference the previous libraries. Place only one referring to your "composite" lib in one of your main include files.
Mircea
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
