AVR Freaks

He's a bit of an idiot if he thinks people aren't going to discuss IDE technologies - it's like religion or politics - everyone has strong views about what works and what does not work. (and don't anyone ever mention TI CodeComposer to me!!)

If they start charging for what was previously an "open" project I think the vast majority of folks would just walk away.

 

My "IDE" is GNU Nano editor and GNU Make. I have a Makefile that gets the target name from the directory it's in (for example "project.cpp" and friends are in a directory named "project". I edit the source and friends with nano, run "make", lather-rinse-repeat. Usually all I have to change in the Makefile is the CPU type and maybe fuse bits or "F_CPU". 

 

Sometimes if I'm super lazy, I use "inotifywait" to remake the executable each time I save an edit. That way I don't even need to go between edit and make repeatedly.

 

Arduino "IDE". Blah!

 

 

All this to write a program to blink the pin 13 led without using delay()... 

 

 

Yeah I get it. On the rare times when I NEED to run Windoze, I always get a "command or filename not found" when I mistakenly type "cp" instead of "copy" or wonder why "cd .." errors or why "nano" doesn't exist or why I can't string a bunch of commands together with "&&" ... ad-nauseaum.​ ​​​​​​

I started with Arduino a few years back after a life event. I needed a distraction, and Arduino was fun, but trying to do my own boards was frustrating, and my C++ skittles are not very colorful, I very much prefer the limitations of C.

 

@Kartman why you are down with R-Pi, it is a Linux networking machine on a $10 board. I have had some running for a few years now. They run the AVR toolchain just like my Ubuntu desktop. But sadly I can not use VScode to remote into them (they are Zeros, not 3 or 4)

 

Oh yes.. agree completely. The way Arduino is set up anyone can develop a cool project and have a very good chance of it working, which results is gratification (positive feedback) that builds on itself. No complaints here.

 

But, those who are encouraged by their success invariably learn more and develop more complex projects.

 

And THAT'S where the problems begin. The Arduino people's desire to not waste a single byte of flash prevents new users from realizing that, yes, CPP does indeed support their ideas and needs, but some of these functions are WITHHELD from them in the name of saving a few bytes. Furthermore, the option to USE these functions at the risk of building a larger (or slower - or both) binary is denied them.

 

They want to save memory? That's OK, but it's MY ARDUINO and if I want to spend more flash, the Arduino people should not be telling me "it won't work' or "it's not supported" or "no, we won't tell you how to do it the standard way".

 

Also, some of the well known people are downright rude, arrogant or condescending. Others pass on and perpetuate plainly incorrect information.

 

Sorry to rant about it, but so often I've spent time writing lenghty posts on "that other site" genuinely trying to help a person who just needs a few things explained and has instead been torn apart by the "experts".

 

I know what it feels like to finally begin to understand something and just need help AT MY LEVEL to finally "get it" and I really hate seeing people on the verge of understanding be instead turned off by a rude know-it-all or, worse, someone who doesn't take the user's LEVEL of understanding into account and swamps them with too much info.

 

I've been told that I'm a "great teacher", but all I really do is try to get students to UNDERSTAND, not just memorize. If a person understands, they have no need to memorize.

 

 

 

Yes, "float" and "double" are (in Arduino land) the same thing. I understand the importance of conserving resources in a small embedded microcontroller environment, but the OPTION to use doubles should be available for those who need it (or merely want it). Same goes for floating point output and printf. Leave them out by default, but give the user the option to use them if he/she wants, rather than withholding it from the user because, of course, THEY know better than a mere mortal beginner! 

 

The AVR toolchain (avr-gcc / avr-libc) has a nice function to setup any character device (serial, LCD, etc) to use stdout, stderr and stdin if appropriate (no stdin for an LCD display obviously!) 

 

I wrote a simp!e library for the Arduino that handles all the stuff involved in connecting streams to stdin/out/err with a simple object. For examp!e, to connect stdin/out/err to a serial port, all you do is include the tiny library and write "STDIO.open (Serial);"

 

It can even connect different devices at the same time (for example stdin and stdout to serial but route stderr to something else like another serial port or an LCD). I use it all the time to send debug data to  VFD display for testing.

 

It's handy to be able to use printf and formatting rather than the clumsy bunch of Serial.print calls. Without printf, even controlling numeric output is a pain (imagine calculating a digit's size and then sending the proper number of leading spaces to keep a number aligned on the display rather than simply using something like "%5u" to keep an int lined up!). Or setting up a buffer, then using "dstrtof" merely to print a float value rather than simply saying "%7.2f" (that is, IF they were nice enough to tell the newbie how to link in the float runtime so that the dreaded "?" wasn't displayed instead of the number!). And, "dstrtof" doesn't even provide the user defined number of digits after the decimal point like printf does.

 

All.this baloney to save 1 or 2 kb of flash when the binary barely uses 10% of the progmem anyway! Do they realize that all the extra, clumsy code required to do it the "Arduino way" takes extra space as well?

 

"Don't use printf, instead use a dozen Serial.print calls and a digit size calculation function. It will save space and make the code run 3 microseconds faster".

 

Pardon me while I roll my eyes and slowly shake my head...

 

And if someone dares to use malloc()... oh boy the flames really light up!

 

I hate being kicked in the face for trying to help. That's why I don't go "there" any longer.

 

Ugh... sorry for the rant again. I need to let it go...

 

 

Thank goodness that never happens here... I think you've missed Kartman's point.  This is not Arduino's 'fault'.  It is AVR GCC's 'fault'.  There are other toolchains which (e.g. ICC) which support 64-bit wide doubles, but they are not free and not open.  That won't fit the Arduino mold.  There is partial and experimental support for it in the very latest version of AVR GCC, and if it ever gets flushed out then it will also be available for Arduino.

 

 

There is another angle to Arduino, it sort of started with Hernando Barragan's work, and has a history.

 

https://arduinohistory.github.io/

 

For me, that leaves a bad taste. Wiring software itself is now days licensed for non-commercial use only (form memory, last I looked).

 

 

Wow. That was incredibly interesting. I had no idea what happened behind the scenes (or rather, I had the completely WRONG idea). Thanks for the link. Verrrry interesting..

 

Interestingly, I never really liked the "Arduino way". It looked to me to be an attempt to mold plain old C programming into kindergarten version of Java in an attempt at being "(new) user friendly".

 

I actually go out of my way to avoid non standard stuff like "digitalWrite" and use things like "PORTB |= (1<<3);". The Arduino specific functions always seemed like a crutch to me. I wanted to stay independent of them. The "Arduino way" always felt "dumbed down" to me (if that makes any sense?).

 

After reading the info you pointed me to, I now think I understand where my feelings came from (not the player's dishonesty, but the Wiring and Processing heritage that Arduino has).

 

 

 

Speaking of Arduino options, here is my "Preferences" window of my hacked Arduino "IDE":

 

(click thumb for full size)

 

Probably the handiest options are the floating point options.