Use SetDatabaseDouble() for an integer?

[the.weavster]

I will be receiving some SQLite database query parameters in JSON format but of course JSON only has a generic Number data type. I have experimented just using SetDatabaseDouble() every time and it seems to work but is it working by luck?

Is a Double with no fractional part the same as an Integer?

I know I could adapt the JSON to include a member that denotes the type but I'd rather avoid doing that if SetDatabaseDouble() will work for integers too (Double and Integer are the only number types I need).

[NicTheQuick]

It's not always the best solution because integers are more precise in the integers alone. In the code you can see that two different integer values result in the same double value.

Code: Select all

Define i.i, d.d
i = $7fffffffffffffff
d = i

Debug i
Debug d

i = $7ffffffffffffe00
d = i

Debug i
Debug d

[the.weavster]

Thanks @NicTheQuick

I guess this feature request has just become more important to me

[spikey]

It depends how big an int you need, NicTheQuick's examples both massively overflow the specification. It's not really surprising that it breaks down. The int part of the spec allows 52 bits not including the sign. In fact stackoverflow tells me that it should work fine all the way up to 2^53 because the exponent is still small(!)... See https://stackoverflow.com/questions/184 ... n-a-double
If you can remain within the limits of the spec, and check for out of bounds input data in your program, you should be ok because the int can be represented internally properly.

Code: Select all

Define i.i, d.d

; 2^52
i = 4503599627370496
d = i

Debug i
Debug d

; 2^53
i = 9007199254740992
d = i

Debug i
Debug d

; 2^53 + 1 is the first point of failure.
i = 9007199254740993
d = i

Debug i
Debug d

[the.weavster]

Thanks @spikey

2^53 + 1 is the first point of failure.

I think I can live with that