Arduino Uno: Missing Libraries

[James AlfordParticipant @jamesalford67616]

A couple of years ago, I built a circuit to operate a clock, the circuit being triggered by events generated by an Arduino Uno. The circuit included a RTC and a LCD to display the time and three buttons to set the time. The thing worked perfectly for several days, so I unplugged everything and put it away.

I pulled it out today, reassembled it, plugged it in and the thing seems to have died. The LCD displays a random time initially, but then starts to act like a counter, changing time rapidly more than once a second. None of the time setting buttons worked either. I checked the connections and all are sound. I also changed the battery in the RTC. The board is powered by a transformer.

Suspecting that the code had become corrupted on the board, I tried to recompile the source code and reinstall it onto the board. However, when I tried to verify the code before uploading it, it failed, telling me that the following libraries are missing: event.h, timer.h and elapsedMillis.h.

I tried to reinstall these, but they do not appear in the list of available libraries and I have no idea where I found them originally.

Can anyone suggest a way around this, other than starting all over again?

Thank you.

James.

[John HaineParticipant @johnhaine32865]

You could try putting  #include arduino.h right at the start of your source code, also perhaps reinstall the Arduino IDE which should bring a full set of libraries with it.

[Stuart Smith 5Participant @stuartsmith5]

Have you tried to install via the ‘manage libraries’ option within the tools section of the Arduino IDE?

Other than that ,you could try reinstalling the Arduino IDE.

 

Stuart

[James AlfordParticipant @jamesalford67616]

Thank you both for the suggestions which I shall try in the morning.

James.

[James AlfordParticipant @jamesalford67616]

Oh, well. Back to the drawing board. I tried these and recovered the elapsedMillis library, but the others remain elusive. I suspect that they came with some timer codes that I copied from an on-line project on which I based my code. Sadly,  the site from which I copied them is no longer working.

Regards,

James.

[SillyOldDufferModerator @sillyoldduffer]

Before rushing to do a re-write, search the computer for the missing files.   Strictly speaking they are header files rather than libraries, but no need to split hairs!

The problem may simply be that they are in the wrong place and the IDE’s compiler isn’t looking for them in those directories.   (Directory aka folder.)

In C/C++  statements like #include <timer.h>  are common.  The <> brackets mean “look for this file in the compilers list of system directories”.   This is where installed external libraries and processor headers should be.   There are thousands of them.

Alternatively, .h files written locally by a programmer are stored in a list of user defined directories.   They are identified by the quote marked form: #include “timer.h”    For simplicity In the Arduino IDE, local headers like “timer.h” should always be in the same directory as the .ino file.   The editor should open each in a tab.

 

How to Search

In Linux or mac, I prefer to search from the command line with:  find ~ -name ‘event.h’ -print 2> /dev/null

Or “find ~ -name ‘*.h’ -print 2> /dev/null”    note the * wildcard will list all header files.

In Windows cmd.exe offers the less powerful dir command, so it may be easier to use file explorer.  Start explorer and navigate to your home directory.   Then type ctrl-s  or ctrl-f to open the search box.   Type the filename into the box and with luck Explorer will find it.  Fingers crossed, the missing header files are only misplaced and can be copied to the right directory, or their content pasted into a new file.

Cause.  I guess the missing header files are the  #include “something.h” type?  If so, not often necessary on Arduino, making it easy to forget they exist.  If the project is copied later, say going from version 1 to version 2, or moving to a different machine, only the .ino is taken and the equally important .h files get left behind.  You can guess how I know!

Dave

 

[James AlfordParticipant @jamesalford67616]

Thank you for the suggestion, Dave.

I have just had a quick search and found them lurking in an old folder, date from 2020, which is encouraging. I shall try to move them to the expected location this evening.

James.

[James Alford]

On James Alford Said:

Oh, well. Back to the drawing board. I tried these and recovered the elapsedMillis library, but the others remain elusive. I suspect that they came with some timer codes that I copied from an on-line project on which I based my code. Sadly,  the site from which I copied them is no longer working.

Regards,

James.

Try putting the defunct website url into the internet archive Wayback Machine.
https://web.archive.org/

They may have captured it.

Robert.

[James AlfordParticipant @jamesalford67616]

Thank you, Rupert.

I knew that there was something like this, but I didn’t know its name. I shall have a look.

[James AlfordParticipant @jamesalford67616]

Well, good news and bad news.

The good news: I found an archived copy of the website that I used.

The bad news: sadly, the page with the code and libraries had not been archived; only the pages describing the code.

The good news: I found the missing libraries, lurking in an old folder as suggested, and copied them into the current location.

The bad news: the code still fails to compile, saying that the .h file is missing. Forcing the compiler to skip this only finds further, unrelated errors.

I do not know quite what had happened, but the combination of a new laptop and reinstalling the IDE on the new computer when I bought it seems to have caused major problems with code that worked previously.

Ho, hum. Think again.

Thank you all for the help.

Regards,

James.

 

 

 

 

 

[James Alford]

On James Alford Said:

…
The good news: I found the missing libraries, lurking in an old folder as suggested, and copied them into the current location.

The bad news: the code still fails to compile, saying that the .h file is missing. Forcing the compiler to skip this only finds further, unrelated errors.

Can you post a screenshot of the directory listing, and where is the ‘current location’? Also a pic of the source code showing the #include statements.  I ask because the compiler is very fussy!

Quite likely the unrelated errors are related to the missing header file – programs depend on the foundations being all present and correct.

I do not know quite what had happened, but the combination of a new laptop and reinstalling the IDE on the new computer when I bought it seems to have caused major problems with code that worked previously.

Ho, hum. Think again.

If all else fails, at least you can print them out, sob.  Moving a multi-file program to a new is a bit tricky – the IDE  expects a disciplined structure and a copy is easy to mess up.

 

Thank you all for the help.

Regards,

James.

 

 

 

 

 

The problem is probably trivial, it’s just nailing exactly what it is that’s difficult!

🙁

Dave

[James AlfordParticipant @jamesalford67616]

Thank you, Dave.

This is where the libraries and the ino file sit:

These are the libraries: the event.h file is within the Timer-master folder.

 

This is the code.

Regards,

James.

[John HaineParticipant @johnhaine32865]

Looking back at your first post, the problem could be that the RTC (ds3231?) had lost its configuration when you changed the battery.  IIRC these have a wide range of programmable output frequency options so when powered up in its default state it may not be at the frequency you expect.

[SillyOldDufferModerator @sillyoldduffer]

Hi James, the screenshots have me flummoxed because they look correct.  Appears you’ve got the file and directory locations right, which is good, except it still doesn’t work!

I’ll have a look at how the Arduino IDE adds libraries to the compiler’s search list.   Might be done by the installer, in which case moving the libraries manually may may have missed a step.  May take a while, I’m unwell this morning.

If you fancy a go yourself, try setting the “Show verbose output during compile” tickbox, which is in  File->Preferences->Settings.   After that Verify the sketch and wade through the output looking for the library loads.   Comparing a successful #include with a unsuccessful #include may reveal the answer.   Unfortunately the log is huge, and written in techno-terse!   Look for lines containing the word library.

Might give the answer, but I have more hope that an internet search for “manually install Arduino library” will provide the answer – the missing step.

Just in case anyone reading this thinks Arduino is always this complicated, not so.  James has an unusual problem that’s forced us to open the bonnet and take the engine apart…    Rarely necessary!

Dave

 

[James AlfordParticipant @jamesalford67616]

Thank you John and Dave.

I shall look at both of your suggestions, realistically tomorrow as we are off out shortly.

I hope that you feel better soon, Dave.

Regards,

James

[James AlfordParticipant @jamesalford67616]

Hello, again.

I did some more investigation into the problem with my code, reinstalled the IDE, reverted to the version of the IDE that I used to write the code and searched again for the library that was installed, but that the compiler could not find. Still, frustratingly, to no avail.

My son suggested pasting the code into ChatGPT to see whether that could find the problem. It found no errors in the code, but made suggestions, similar to those made on here. Again, to no avail.

I then wrote my requirements long-hand and pasted them into ChatGPT. A few seconds later, it had written the code, explained how to connect the components required to make it work and given an explanation of how the code works. The code then complied seamlessly.

I was impressed, but frustrated that I could not find the cause of the problem.

Thank you for the help given here.

James.

[Charles LamontParticipant @charleslamont71117]

So what were the necessary alterations (if not too many)?

[James Alford]

On James Alford Said:…

I then wrote my requirements long-hand and pasted them into ChatGPT. A few seconds later, it had written the code, explained how to connect the components required to make it work and given an explanation of how the code works. The code then complied seamlessly.

I was impressed, but frustrated that I could not find the cause of the problem.

…

This news should have us all trembling in our boots.  As far as I  know this is the first time that an Artificial Intelligence has beaten this forum to the answer!

Thanks for letting me know – had to put the problem aside due to poor concentration earlier and was gritting my teeth for another go.

Dave

[Charles Lamont]

On Charles Lamont Said:

So what were the necessary alterations (if not too many)?

To be honest, I am not sure. Mainly, the new code does not use the event.h library. The rest bears a resemblance to my original code in places, although some sections are far more elegant than mine.

[SillyOldDuffer]

On SillyOldDuffer Said:

On James Alford Said:…

I then wrote my requirements long-hand and pasted them into ChatGPT. A few seconds later, it had written the code, explained how to connect the components required to make it work and given an explanation of how the code works. The code then complied seamlessly.

I was impressed, but frustrated that I could not find the cause of the problem.

…

This news should have us all trembling in our boots.  As far as I  know this is the first time that an Artificial Intelligence has beaten this forum to the answer!

Thanks for letting me know – had to put the problem aside due to poor concentration earlier and was gritting my teeth for another go.

Dave

I was just surprised at how quickly ChatGPT generates the code and converts natural language into code, allowing tweaks to be made very quickly. The downside is that if it goes wrong, I shall have even less chance of finding out why.

[James AlfordParticipant @jamesalford67616]

A bit of an update on my experience with ChatGPT.

As mentioned, ChatGPT wrote code for my requirements extremely quickly and, when loaded into the IDE, compiled. However, it did not actually work in the way that I wanted on the Uno itself. This highlighted the importance of being explicitly clear about the requirements of the application for which the code is intended.

Another frustration is the inconsistency of the output from ChatGPT. I gave it a set of requirements from which it generated code. The code failed to compile in the IDE. I then gave ChatGPT exactly the same requirements with no changes at all to the script that I fed it. This time, it produced code that worked.

It does not always find all of your requirements either. I explicitly wrote for it to not use a particular function in the code as it always failed to compile. Sometimes, it would produce code without the function; other times it would ignore my requirement and use the function.

I also tried Copilot on Edge which works in a similar way and has similar foibles.

Overall, these two tools are a great help for writing code quickly from plain English, but they require a lot of patience and a lot of testing.

James.

[IanTParticipant @iant]

I’ve been watching your progress with some interest James.

I’m in the process of trying to build a new remote display for my chinese mill DROs. The remote display (I originally purchased) has a small LCD display that’s hard to see and (less than great) membrane buttons. I’m planning to replace it with a 17″ Monitor and I already have a prototype ‘XYZ’ display up and running. I’m using a PicoMite VGA controller to do the donkey work.

Without boring everyone with the details, I’m now busy trying to decode the DRO O/P signals and believe they are either BIN6 or 21Bit. I’m leaning towards 21Bit after some poking around with a scope last night but I’m not entirely certain yet.

Welcoming anythng that would give me a head start, I found some Arduino code for BIN6 scales that looked useful but needed it in MMB. I was wrapping up last night but (on a whim) I posted the Arduino BIN6 code into CoPilot and asked it to translate it into MM Basic.

I expected CoPilot to know something about Arduino but didn’t really expect it too understand MMB. I was really suprised to immediately get a clean Arduino listing back, plus what looks like a very good MMB routine too. I haven’t tested the MMB code yet (and may not do so now) but CP has already saved me a good deal of time in trying to manually translate the (working) Arduino code vs starting from scratch. I’ve not used Arduinos for about 10 years now, whereas MMB is very familiar to me now, so having CP get me even half way there, is a big deal for me.

I also think that AI is probably going to be very disruptive once it really gains traction…

Regards,

 

IanT

[James AlfordParticipant @jamesalford67616]

I have another update on my experience with Copilot and ChatGPT; along with a couple of follow-up questions.

I have finally succeeded in recreating code for the clock that I should to build, one day. I used ChatGPT initially as it seemed to be working well. However, when I studied the code that it had created, it was full of pin conflicts and other problems. No matter how I rephrased my requirements, I could not prevent the conflicts. I then tried Copilot which, after a number of false starts and several hours work, produced working code. You need to be extremely clear with your instructions, yet not too rigorous; an odd requirement. When I was very prescriptive the code would not even compile. Allowing the software more flexibility gave better results.

However, I have a couple of questions.

Question 1

The set up uses a separate RTC to keep time and a LCD to display the time and day. I have buttons to advance and retard the time. Using these I can alter the time displayed on the LCD. I cannot work out, though, whether I need a specific “set” button for the time. Some clocks have them, whereas others do not.

Is a set button needed?

Question 2

The code also sends signals to various pins to flash LEDs via a transistor circuit. There are pins to flash LEDs: once a second, once a minute, once an hour and once a day. It flashes another LED once, twice, thrice or four times on the quarters; and another on the hour, once, twice, thrice or more depending upon the hour.

However, the code runs sequentially meaning the seconds LED stops flashing whilst the minute LED flashes; they both stop whilst the hour LED flashes and so on including when the quarters strike and the hours chime.

Is there a way to modify the code so that these events so that they do not interrupt each other? In time, the LEDs will be replaced with relays to drive a visual display. The pauses will affect the accuracy of the mechanics.

I can post examples of the code if it helps.

James.

[James AlfordParticipant @jamesalford67616]

I still have my question 1:

The set up uses a separate RTC to keep time and a LCD to display the time and day. I have buttons to advance and retard the time. Using these I can alter the time displayed on the LCD. I cannot work out, though, whether I need a specific “set” button for the time. Some clocks have them, whereas others do not.

Is a set button needed?

but may have a partial solution to my question about flashing the LEDS at the same time when needed. Copilot put all of the code that is not part of the initial setting up into a single void loop. This includes the code for setting the time and for each timed function. However, my original code, which I wrote long-hand, had each function in a separate void loop e.g void Second_Trigger(), void Minute_Trigger(), void Hour_Trigger() and so forth. Would splitting my new code into sections like this be likely to help the LEDs to flash at the same time, or at least more closely together?

James.

 

[SillyOldDufferModerator @sillyoldduffer]

An interesting observation about AI and pin conflicts in #775838. I guess the learning behind AI’s suggestion is based on human written clock code that makes the same mistake, which is copied.  Investigating how and why AI gets stuff wrong could be a whole new hobby.

Whether a set button is needed or not question is just ergonomics I think.  Code can be written such that the RTC just counts from wherever the advance and retard buttons got to.  Except, I, being human, am happier with a design that sets the time in a way that positively tells me it’s been done.   As a good interface should never leave the user guessing, I’d have a set button!

Flashing LEDs together is a synchronisation problem.   The answer to ‘Would splitting my new code into sections like this be likely to help the LEDs to flash at the same time, or at least more closely together?‘ is no.   It’s the other way round – a combined flasher that meets the requirement: ‘There are pins to flash LEDs: once a second, once a minute, once an hour and once a day. It flashes another LED once, twice, thrice or four times on the quarters; and another on the hour, once, twice, thrice or more depending upon the hour.‘

The most accurate way is with one of the Arduino’s built-in hardware timers and an interrupt, but I think simpler to write ordinary code inside void loop(), and accept flashes could be wrong in the millisecond region.

The key is to not use the delay() function, or anything else that waits.  delay() stops the program entirely, bad when we need the computer to handle seconds, minutes, hours, and quarters seamlessly.  Instead, the flash is timed with millis(), and which LEDs are switched is controlled by setting a variable.  This code establishes flash duration by testing millis() and doesn’t stop.

Clear as mud I expect, an example is needed!  Off to a hospital appointment now, and then a haircut, but I’ll rough something out in the waiting room and report later.

Dave

[Author]

Posts