I've come across something I can't explain. I am on a VS2019 platform using the latest Visual Micro extension and programming a genuine Adruino Nano (32K flash). 

I've reached the limits of the flash memory available. When compiled, the code equates to 30,486 byes used of 30,720 available. In one section of my code I have a "switch" statement. If I add another switch condition and the below lines of code, the compiled size jumps to 31,180 bytes therefore the VM "too big" message appears. Even if I remove the Serial.println statement (blank between case and break), I still get a compile size of 31,156.

case 'X':
     Serial.println(EEPROM.readByte(Addr_MCnum1));
break;

Has anyone come across this or can anyone explain why this is happening?

b

Thank you for the quick reply!

I can confirm the configuration is set to Release.

I have attached the two verbose compile outputs in one text file. The output without the switch statement will be listed first. The output with the switch statement starts after "==========" separator.

A switch statement is a difficult thing to optimize. The optimal implementation depends not only on the number of cases in the statement but also on their values. The compiler might use a series of conditional jumps if there are few cases only (basically using a series of if-else-else). If the number of cases grows, a lookup table is the better solution (a table that contains the target address for each case). This is faster, but has the drawback that also empty cases need handling (e.g if you have cases "A", "B", and "D", the table still needs to have an entry for "C"). So it is possible that due to the single added case, the compiler switches to using a (larger) table, causing this extra memory requirement. Even if the compiler is told to optimize for code size, it will at one point start trading off size for speed.