Conversion warning in hashed_index.hpp

[mbradle@…]

Hello,

I get a warning that conversion to 'float' from 'size_type' may alter its value when compiling the code ex.13.7.cpp (based on an example from The Boost C++ Libraries by Boris Schaeling) with -Wconversion. The attached file test.txt shows the compilation. The attached file diff.txt shows a change that fixed the problem for me. Thanks, and best wishes.

Brad Meyer

[mbradle@…]

test cpp file

[mbradle@…]

transcript of compilation and warning

[mbradle@…]

Change that removed the warning (possible patch)

[Joaquín M López Muñoz]

Hi Brad,

Actually, your proposed patch is not entirely correct: n should be promoted to a floating point type, not mlf to integral. Please try the following:

      float     fbc=static_cast<float>(1+static_cast<double>(n)/mlf);

If this clears the warning I'll commit the change to the repository. Thank you.

[mbradle@…]

Replying to joaquin:

Actually, your proposed patch is not entirely correct: n should be promoted to a floating point type, not mlf to integral. Please try the following:

      float     fbc=static_cast<float>(1+static_cast<double>(n)/mlf);

If this clears the warning I'll commit the change to the repository. Thank you.

Hi,

Yes, this does clear the warning, and I do see that n should be promoted, not mlf. Thanks, and best wishes.

Brad

[Joaquín M López Muñoz]

(In [83058]) fixed #8129