Re[2]: [lmi] confused by latest expm1.c changes

[Vadim Zeitlin]

On Mon, 23 Jun 2008 22:48:49 +0000 Greg Chicares <address@hidden> wrote:

GC> Still, if we can avoid Yet Another Make Alternative, then we should.
GC> I know, in theory, you cross your fingers and everything just works;
GC> but in practice, when it breaks, I have to fix it.

 FWIW I don't like YAMA in general and I have especial dislike for Jam
which doesn't seem to have any redeeming features at all so you don't need
to convince me about the benefits of avoiding it.

GC> > GC> Do you think it would be hard to autotoolize boost::filesystem so that
GC> > GC> we could use it as a shared library?
GC> > 
GC> >  No, it wouldn't.
GC> 
GC> That's good to hear. A makefile along the lines of
GC>   boost/libs/regex/build/gcc.mak
GC> would be quite sufficient, if that's easier. In fact, if you want
GC> to work on that now, I'd be glad to have it, though I'm in no rush.

 I'm afraid I can't do this right now, I'm still not totally done with MSVC
(I'm reviewing the patch and trying to minimize it/make it more palatable)
and I'd really, really like to finish this as it's hanging over me since
several months already. And once the main MSVC patch is accepted I'd like
to work on PCH support (for it and maybe g++ under Linux too) to make the
compilation faster as I waste too much time currently waiting for LMI to
build... But I'm putting this in my TODO list.

GC> Let's think this through.
[...snipped because I agree with all this and don't have anything to add...]

GC> >  So everything being said I think it could be simpler to just copy the 3
GC> > files boost/math/special_functions/{log1p,expm1}.hpp and 
details/series.hpp
GC> > to LMI source tree and simply use them.
GC> > 
GC> >  What do you think?
GC> 
GC> Have you tried that, just to make sure they can truly stand alone;
GC> or can you easily remove any dependencies they might have?

 No, sorry, I didn't do it yet. But it doesn't look like they have any from
reading the code and I will try it soon.

GC> Do they have expm1l() and log1pl() with the '-l' suffix?

 They don't but their log1p() is a template function so it can be used with
any of float, double and long double so it's not really needed. They do use
expm1l() and log1pl() to define long double overloads if they're available.

 Regards,
VZ