Why is float's default precision only 2 decimal places?

[tremby]

At https://github.com/laravel/framework/blob/master/src/Illuminate/Database/Schema/Blueprint.php#L505 the default precision of float is set to 8 digits, 2 max after the decimal place. This is absolutely bizarre. That's the kind of thing decimal is used for. When I want a float I want a huge amount of precision (or double for even more of course), and I see no reason for the framework to presume 2 decimal places is enough.

[taylorotwell]

Just pass the amount of precision you want. No matter what default I pick, someone will want to change it.

[tremby]

What's wrong with just float with no precision arguments? See the MySQL docs at http://dev.mysql.com/doc/refman/5.0/en/floating-point-types.html

MySQL permits a nonstandard syntax: FLOAT(M,D) or REAL(M,D) or DOUBLE PRECISION(M,D). Here, “(M,D)” means than values can be stored with up to M digits in total, of which D digits may be after the decimal point.

and

For maximum portability, code requiring storage of approximate numeric data values should use FLOAT or DOUBLE PRECISION with no specification of precision or number of digits."

The former, which as far as I can tell in Laravel is the only kind of float I can make with the schema builder, is non-standard. Seems odd to only allow this. And I could be wrong about this, but doesn't a float set up in MySQL like this behave exactly like DECIMAL(x,y)? If that's what the user wants the user should be using decimal.

The latter (FLOAT with no specification of precision or number of digits) is as you see above what is recommended, and it's exactly what I want to use, and I imagine what most users would want to use in most cases. But am I right that there's no way to specify this with schema builder at present? Or how is it done?

And I haven't mentioned the third possibility mentioned in the docs, where you pass a single argument to float to give the precision (which is not the same as either argument of the FLOAT(M,D) syntax. As far as I can tell there's no way to specify this with schema builder either.

So right now as far as I can tell the only way to set up a column for an actual floating point number is to make a fixed precision "float", and then run a non-portable, database-type-specific statement to change that column to be a proper float.

But honestly my main beef here is that I simply can't think why any developer would be looking for a float and say "yup, that's exactly as I expected" when they see it is limited to two decimal places. That's just not what a float is.

Thanks.

[chrisfranko]

I just ran into this same problem... I need big float numbers but when I noticed it was cutting it off to 2 after the decimal.. that isnt even a float lol

[fisharebest]

Can I ask you to reconsider closing this. The fix is just to copy a few lines from double() to float(). This would allow us to use ->float('col', null, null) in the same way we can currently use ->double('col', null, null).

From class MySqlGrammar...

protected function typeFloat(Fluent $column)
{
    return "float({$column->total}, {$column->places})";
}

protected function typeDouble(Fluent $column)
{
    if ($column->total && $column->places)
    {
        return "double({$column->total}, {$column->places})";
    }
    else
    {
        return 'double';
    }
}

Surely the inconsitency between the two implementations must irk you as much as it irks me ;-)

If you don't have the time, I'm happy to create a PR.

[iamgoodbytes]

Is this case still open? I tried changing float length from 2 to 8 positions accuracy in order to be able to store latitude and longitude values according to googles mysql guidelines, but float doesn't seem to support these length parameters in schema builder

[browncolyn]

Let's just assume that for some reason doubles are always better than floats. If I want a double, I would have asked for one. Why is laravel forcing me to use a double? It's like going into McDonalds and asking for a hamburger but give you a salad because its better for you. Can't I just get a float? What's the rationale here?

[jdavidbakr]

+1 - if you need a double, you have a double, but float should always imply the float type. Yes, there are dangers with using the float if you don't know what you are doing, but there are cases where you want to use the float when you do, and it is annoying to have to create a raw query in the migration when you have the float migration type right there. Not sure how it can be fixed since it would break backwards compatibility, though.

[newms87]

How is this still an issue? Float is float, double is double, they're different. MySQL obviously recognizes that cause they made 2 different types that behave in 2 different ways. So why make float a double?

[tremby]

How is this still an issue? Float is float, double is double, they're different. MySQL obviously recognizes that cause they made 2 different types that behave in 2 different ways. So why make float a double?

Laravel isn't "making a float a double". I think perhaps "double" dosen't mean what you think it does.

[fisharebest]

Laravel isn't "making a float a double"

Yes it is :-(

When this issue was created (laravel 4), floats were created as decimals. (See my post above)

	protected function typeFloat(Fluent $column)
	{
		return "float({$column->total}, {$column->places})";
	}

But since Laravel 5.0, it creates them as doubles:

    protected function typeFloat(Fluent $column)
    {
        return $this->typeDouble($column);
    }

The defintion of double has not changed.

    protected function typeDouble(Fluent $column)
    {
        if ($column->total && $column->places) {
            return "double({$column->total}, {$column->places})";
        }
        return 'double';
    }

[tremby]

Where is that code?

The code I linked to when I opened this ticket (haha, four years ago!) doesn't look it has changed: it still has the same default arguments of 8 and 2.

If in some deeper part of code these arguments are ignored and a double is being used, I'd consider this closed. As long as the default doesn't only have two decimal places of precision...

[fisharebest]

Where is that code?

https://github.com/laravel/framework/blob/5.5/src/Illuminate/Database/Schema/Grammars/MySqlGrammar.php

[tremby]

Oh, so nothing has changed. Whether it's float(m, d) or double(m, d), it's still a fixed number of digits total and a fixed number of digits after the decimal point. Can null be passed instead to use mysql's default? (i.e. double() rather than double(m, d))

[fisharebest]

As long as the default doesn't only have two decimal places of precision

The defaults have not changed.

To get an IEEE double, you must specify a precision of 0 - i.e. $table->double('foo', 0);

There is still no way to get an IEEE float.