20 minutes compilation time for ESP32

Then nobody can help you.

@sonofcy (post42), do you realise that @aiggya was replying to your message about the antivirus and not to your message about the logs?

Thanks, NOPE, my bad. How would I know that, I don't want to make that mistake again.

The reason for the verbose logs is to make 100% sure some form of non local file storage is or is not happening. Even something as forgetfull as you assigned some form of 'temp' work area to a NAS for some reason.
I personally use Msft OneDrive, but I have it configured to ALWAYS store the files locally. In this mode (uncommon) it simply acts as a automatic continuous cloud backup. IOW, the sketches are always on my local SDD (It's a MAC)
So let's have a look so we can rule that out or not.

Am I missing something, is there a setting that displays these messages in what I call a threaded or conversation style ike many forums I have seen so the context is always clear?

It is a bit subtle. This is the indicator that a reply was made in response to a specific previous post in the topic:

If you click on that, it will expand to show the post to which it is responding:

You can click the "Jump to post" link there to see the responded post in its original context in the topic.

@aiggya As a test, try disconnecting the PC from the internet. If the compile works, then it is not an internet-related problem. If the compile fails, why is the internet needed? I just tried it and did NOT need the internet.

Internet disconnected, compilation SUCCEDESS, still slow.

Is there any chance you are using Dropbox, OneDrive, Google Drive?

You sound like my mother.

Click the little arrow
image

and you can see it

You can close it again with the little down arrow at the left or go to the post (right).

Should I be flattered? Not sure your Mom would be happy being compared to a man though.
We still have the outstanding question of 'are you using any form of cloud'

Since of course i can't be sure about what the OP is using, but i have always completely removed any form of build in cloud storage from my OS and have not added any form of that. (It takes a bit of work to remove ((dropbox)) sorry i mean OneDrive ! from Windows) and i have similar long compiles at times.

Just to add some (more?) benchmarks to this discussion, I'm using Arduino IDE 2.3.6 with ESP32 Core 3.3.0 and am compiling for the board ESP32 Dev Module using code from examples > ESP32 > camera > CameraWebServer. I chose this because the test is easy to set up without any hardware required and without necessitating the installation of multiple prerequisite libraries.

With Windows 11 / 40GB memory / i5-1235U processor and local disk only ~3 years old, I get the following times:

First "Verify" (or following rename) : 52 seconds
subsequent "Verify" : 14 seconds

This is what I see in the task manager when it is running:

I can also try the same thing with a Windows 10 device (~10 years old) if anyone is interested in pursuing this

When doing screenshots of Task Manager, expand the node for the IDE to see the individual back-end tasks the app is running.

Also, you need to consider how many (logical) cores you have. On Windows, 100% CPU means all cores maxed out, whereas on Linux and Mac, it's up to 100% per-core.

The i5-1235U is complicated

  • 8 efficient cores
  • 2 performance cores
    • with hyper-threading, so two logical cores each

Totaling 12. So on that screenshot above, 5.4% could be 65% utilization of one core. (On Linux/Mac, that would be listed as 65%. Two such cores would be 130%; as would five cores each at 26% -- it gets complicated.) If you have 20 cores, maybe one is pinned (5%) running a back-end build task and the other 0.4% is the IDE using another core to do something light.

The fact that you don't have at least one core pinned means that -- at the moment the screenshot was taken and given the interval at which the Task Manager updates -- the IDE was not that busy. Also, if the CPU is waiting on I/O, for the disk or network, it is not being "utilized".

Sorry. Yes it was a bit crude and I know that it was not an ideal basis for comparison, especially across OS types. It was more to show that there was not simultaneously a bit coin miner or similar running.

Is this closer to what you mean ?

or are there better statistics which allow a more meaningful comparison for attempting to see if users, who suffer performance problems with the IDE, have configuration or other issues ?

Probably.

I dont use any form of cloud, disconnected internets and the compile time was not altered.

Then you need to post the VERBOSE compile log so somebody can analyze it and see where it slows down.

How would you see that? It’s not that there is any timestamps?

Ok, I thought there were but I don't even see any setting in the Command Palette.

Maybe someone else knows how to get timestamps?