Date: Mon, 17 Nov 2008 09:10:46 +0100 From: Daniel Kr?gler Subject: Re: [Boost-users] timeconv.inl To: boost-users@lists.boost.org Message-ID: Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Dan Smithers wrote:

...

...

I am porting some inherited code to MSVS 2008 This is also bringing along an upgrade from boost 1.32 to a newer version (probably 1.37).

I have found a thread wrapper class in the code that uses time_to directly. This is defined in lib/thread/src/timeconv.inl and doesn't look like its meant to be used directly.

I agree with the last assertion. I'm not sure that I correctly understand with "I have found a thread wrapper class in the code that uses time_to directly"?! In which code or in which wrapper class?

In the inherited code.

...

...

The original code base has timeconv.inl in the boost include directory.

Was timeconv.inl moved at some point in the past?

Is this as bad as I think it is?

I don't see the (bad) point - timeconv.inl was already part of the implementation of boost 1.32, so it's existence or it's precise location shouldn't influence your code [Note that the helper functions are simply wrapped into an unnamed namespace w/o further surrounding namespace].

It's now located in duplicated form in lib/thread/src/win32 and lib/thread/src/pthread. Whether this is intended, I don't know, but it also shouldn't harm.

Thanks, I looked around a bit more, and found that the example code used the same code as in timeconv and that I was making exactly the same set of calls to achieve the same result. It just doesn't seem right to be including directly from the src directory. In the code base I have inherited, timeconv.inl is in the include directory and appears to be for public use.