APFloat::convertToDouble asserts

Project Infrastructure LLVM Dev List Archives
1

Hi!

I wonder why APFloat::convertToDouble asserts if it is not an
APFloat::IEEEdouble while the name "convertToDouble" suggests that it
converts when it is not a double.

-Jochen

2

The conversion is to (host) double, as the name implies. Among APFloat types, that is only implemented for APFloat::IEEEdouble.

3

I wonder why APFloat::convertToDouble asserts if it is not an
APFloat::IEEEdouble while the name "convertToDouble" suggests that it
converts when it is not a double.

The conversion is to (host) double, as the name implies. Among APFloat types, that is only implemented for APFloat::IEEEdouble.

yes, it's to host double ("convert this APFloat to double"), therefore I hoped that it works for any APFloat type (otherwise I would suggest a name such as getIfIEEEdouble() ).
Is there a reason against it or would it be accepted if it would work for any type?

4

I'd rather not. The functionality you want is there, feed another APFloat type through APFloat::convert first.

Using host FP is not something that should be encouraged; the main point of APFloat is so people don't have to do that. Why do you want to, btw?

5

I'd rather not. The functionality you want is there, feed another APFloat type through APFloat::convert first.

Using host FP is not something that should be encouraged; the main point of APFloat is so people don't have to do that. Why do you want to, btw?

Yes, i got it working using APFloat::convert. I need host float to output to my backend which is similar
to the c-backend, but generates if/else instead of goto. I do not use APFloat::toString since it can
generate "1" for 1.0 which is not useful for me since I append "f" for float and "1f" is not a valid float.

-Jochen

6

I'm not sure converting to decimal is the way to go for that. What are you planning to do about NaNs?
Something like this would handle all cases:

union { int x; float f; } u = { 0x7f.... };
u.f

7

Yes, i got it working using APFloat::convert. I need host float to output to my backend which is similar
to the c-backend, but generates if/else instead of goto. I do not use APFloat::toString since it can
generate "1" for 1.0 which is not useful for me since I append "f" for float and "1f" is not a valid float.

I'm not sure converting to decimal is the way to go for that. What are you planning to do about NaNs?
Something like this would handle all cases:

union { int x; float f; } u = { 0x7f.... };
u.f

currently NaN's simply do not occur as I do not allow full featured c++ code as input. I have my own
double to string function which could handle NaN if necessary. it then should output a string like
"std::numeric_limits<double>::quiet_NaN()"