Re: Meaning of iniline in declspec(dllexport)

From:

"David Ching" <dc@remove-this.dcsoft.com>

Newsgroups:

microsoft.public.vc.mfc

Date:

Sat, 31 Mar 2007 23:59:02 GMT

Message-ID:

<aBCPh.4856$Kd3.4593@newssvr27.news.prodigy.net>

"Doug Harrison [MVP]" <dsh@mvps.org> wrote in message
news:ffit035ble02p81mq9gdblrlf4a4o6m86m@4ax.com...

On Sat, 31 Mar 2007 17:20:54 GMT, "David Ching"
<dc@remove-this.dcsoft.com>
wrote:

Well, yeah, if you don't want the function to be inlined, you shouldn't
define it in the class body. You could say that in the age-old battle
between dllimport and inline, inline wins. I don't think it's that big a
deal, at least not for exported classes, because IMO, if a DLL exports
entire classes, clients of that DLL are statically linking to it for
compilation dependency purposes anyway.

Well, no where is it written in stone that for inlining to work, the
function needs to be defined in the header. It's only due to compiler
limits that it (sometimes) is that way. And also because historically,
"inline" was an attempt to get away from multi-line #defines, similar to the
way enum also got away from #define. Anyway, I agree that this discussion
has become rather cerebral at this point. :-)

-- David