RE: [boost] Re: date_time issues

From: Jeff Garland (jeff_at_[hidden])
Date: 2002-10-19 18:43:29

> I have managed to compile the release of my project with date_time without
> linking to the release date_time library. The same project needed to be
> linked to the debug library when compiling as debug (MVS7). Strange.

Not really.  If you don't do any I/O calls you won't need the library.
The library contains a only few methods that access some string, the 
other 95% goes inline.
 
> My comment on warnings would be that preferably libraries should not
> generate any warnings if possible simply so that developers can take note of
> their own warnings when they generate them. However I appreciate the effort
> involved in this library so I should not be too picky.

I agree, or at the very least these should be documented. Anyway, I'm going to 
try and tame some of the warnings (see previous mail). However, it is often 
the case that the author of a boost library doesn't have direct access to 
all or even most of the compilers that users use. So we are dependent on 
user reports, regression tests, and patch suggestions from users of these 
compilers.  I for one don't happen to own VC7 and the regression test didn't
show any warnings. So until someone mentioned it today I wasn't aware of the
warnings...

http://boost.sourceforge.net/regression-logs/cs-win32-diff.html

Jeff