Clamp function

[olafvdspek@…]

A clamp function comes in handy in certain cases. Could one be added to Boost? I'm not sure which component would be most appropriate.

The function may be something like this:

template<class T>
T clamp(T v, T a, T b)
{
    return max(a, min(v, b));
}

[Marshall Clow]

I just added "sandbox/boost/algorithm/clamp.hpp" Tests in "sandbox/libs/algorithm/clamp/test".

Let me know if that's what you wanted.

[OlafvdSpek@…]

Yes, except you put the value at the back instead of at the front, I'm not sure that's preferable.

Would it be better to use separate templated types for lo and hi to avoid the issues with std::max and std::min? The return type is clear in this case, the type of the original value.

[OlafvdSpek@…]

return val < lo ? lo : val < hi ? val : hi;

Should be:

return val < lo ? lo : hi < val ? hi : val;

[Marshall Clow]

I can see putting the value first, that's no problem.

However, I don't see the difference in the two code snippets. When hi == val, the first one returns hi, and the second one returns val. When comparing objects, the first calls val.operator< ( hi ), the second hi.operator< ( val ). So?

[Olaf van der Spek <OlafvdSpek@…>]

There is a difference when hi and val are equivalent but not equal. It's not an issue for simple types but it might be an issue in other cases.

[Marshall Clow]

Updated in the sandbox - revision 54484.

[olafvdspek@…]

Looks good.

[olafvdspek@…]

​https://svn.boost.org/svn/boost/sandbox/boost/algorithm/clamp.hpp

[olafvdspek@…]

To which library will this function be added?

[Marshall Clow]

Boost.Algorithm; once it (the library) is reviewed and accepted.

[anonymous]

This kind of request must be sent to the Boost ML, not the Trac.

I propose to close this ticket as this is more related to new features to the forthcoming Algorithm library.

[Olaf van der Spek]

I propose to close this ticket

Proposing isn't the same as doing... Why don't you reassign to the algorithm lib?

[viboes]

Replying to Olaf van der Spek:

I propose to close this ticket

Proposing isn't the same as doing... Why don't you reassign to the algorithm lib?

There is no Algorithm library. So no need to make defects on libraries that doesn't exists. If you can not assign a component, maybe you should close the ticket yourself.

[anonymous]

It's a feature request, not a bug report.

[viboes]

Replying to anonymous:

It's a feature request, not a bug report.

There is Library Submissions for that, I think.

[anonymous]

No, that's for submitting a library, not for requesting a function.

[viboes]

It is not useful to make a request that nobody can handle :) Please close it yourself.

[anonymous]

marshall appears to be handling it. What's the problem?

[Marshall Clow]

The reason that I assigned this ticket to myself is that I have implemented this in the (proposed) Boost.Algorithms library, which I will post a version 0.1 on the man list "REAL SOON NOW".

[sorokin@…]

So, will this function be included in boost?

[Marshall Clow]

See <​http://lists.boost.org/Archives/boost/2011/03/179561.php>

[Marshall Clow]

Checked into trunk in [76388]. Will close ticket when merged to release (which will be for 1.50 release)

[Marshall Clow]

(In [78557]) Merged changes for Boost.Algorithm to release; Fixes #6596; Fixes #6689; Fixes #3215; Fixes #6840

[Olaf van der Spek <olafvdspek@…>]

Is this supposed to be available in namespace boost or just in boost::algorithm?

[Marshall Clow]

boost::algorithm

All the algorithms in Boost.Algorithm are in that namespace.

[Olaf van der Spek <olafvdspek@…>]

Why's that? String functions, like to_upper, are in boost itself.