Floating point values have a limited precision. Hence a value might
not have the same string representation after any processing. That also
includes writing a floating point value in your script and directly
printing it without any mathematical operations.

If you would like to know more about "floats" and what IEEE
754 is, read this:
http://www.floating-point-gui.de/

Thank you for your interest in PHP.

Mr Info 

Are you sure that code you posted is actually the code that's producing that output?

https://3v4l.org/v9k54

Yes I have posted the exact code.

How is this not a bug?
This is the exact code. I'm rounding with precision, it works on Linux, doesn't work on Windows.

I am on Windows 10 Pro x64 and ran this from command prompt.

C:\Users\Tylae>php -r "echo round(-80.0035326999999938, 6);"
-80.003533

> I am on Windows 10 Pro x64 and ran this from command prompt

yeah, but you didn't mangle the "precision" config (look at the subejct)

garbage in, garbage out and so we are at: Floating point values have a limited precision. Hence a value might not have the same string representation after any processing. That also includes writing a floating point value in your script and directly
printing it without any mathematical operations

php -n -d precision=100  -r "echo round(-80.0035326999999938, 6);"
-80.00353300000000444924808107316493988037109375

Also note that round() returns a float.

Oh I see, our precision on the win box in the php.ini is different... I guess that's the difference we see. Not sure how to fix it now but thanks.

If it's any help at all, the most I've changed in my php.ini file is adding the file path for logging, otherwise everything has been left default. Maybe do a diff with a default php.ini file against your current php.ini file.

tiffany, we actually need a greater precision for scientific calculations. The Linux box is in fact the one that is not properly configured. 
The issue arises when some other values that are not scientific need to be rounded and validated before storage, and this values can't exceed 8 decimals. But round doesn't supersede the precision, which creates the issue.

Any workaround/idea welcome. Some values cannot be stored as string and need to be trimmed (8 decimals max). Those are not the scientific values that require high precision.

@info at globules dot io: Set precision=-1 to display the value with full accuracy but without unnecessary digits.

For fine-grained control use sprintf().

Thanks Nikic :)