Is it worth compiling in math functions by default?

[c9845]

To use math functions like pow() and round(), you currently have to do go build -tags "sqlite_math_functions" .... This works, but this build tag is often forgotten/missed until an error like "not such function" is kicked out.

I think it should be considered to add the math functions into builds by default (aka no build tag required). I think it is common enough to use round(), 'floor(), and ceil() that adding the math functions by default would make sense.

This would help adapting SQL queries from MySQL/MariaDB or Postgres where the math functionality is always present. It would also help swapping in this library versus the modernc/cznic library that has the math functionality enabled by default [1]

[1] https://gitlab.com/cznic/sqlite/-/tags/v1.14.5

[c9845]

So, looking into this, I am pretty sure this is as simple as removing the below lines from the sqlite3_opt_math_functions.go file.

//go:build sqlite_math_functions
// +build sqlite_math_functions

[c9845]

Obviously, the README would need updated to either remove the "Math Functions" from the "Feature/Extension List" table too (or note that the math functions are built-in).

[c9845]

This does not affect round(), but does affect floor() and ceil() as well as everything else from this page [1]

[1] https://www.sqlite.org/lang_mathfunc.html

[dorzepowski]

I would say that we could have cake and eat it too, by setting something like:

//go:build !disable_sqlite_math_functions
// +build !disable_sqlite_math_functions

It will give all those benefits when migrating or supporting multiple db's as mentioned in the issue, while still allowing for decreasing build size in case if someone needs to have smaller binaries and doesn't use those functions.