On Wed, 20 Jun 2001, Nick Ing-Simmons wrote:
> Nathan Torkington <[EMAIL PROTECTED]> writes:
> >Brian Ingerson writes:
> >> Most importantly is acceptance. I don't know how how many of you
> >> have noticed, but there is nearly no cross-posting between the
> >> Inline and XS mailing lists.
> >
> >Until we talked at YAPC, I had three misconceptions:
> > * Inline was for trivial extensions
> > * Inline mandated you put your C code in a heredoc in a Perl module
> > * Inline had one model of compilation: your extension is compiled the
> > first time it's used, and then cached.
> >
> >At YAPC I learned:
> > * Inline exposes almost all the Perl API, so you can do pretty much
> > any XS thing with Inline, only it's easier :-)
> > * Most Inline modules are written with the C code in separate files
> > where it belongs. The heredoc stuff I'd seen was an aberration.
> > * Inline also has an h2xs-like mode, where you make and make install,
> > and it's compiled and installed in much the same way as a regular
> > XS module.
> >
> >I can't speak for others, but that's why I'd always figured Inline was
> >no match for XS. Now I'm a convert.
>
> My understanding was things were "Inline" as well.
> As compiling Tk takes several minutes on a 1GHz machine, and an hour+
> on a slow one it did not seem to be worth looking at.
> If it has a way to build .so/.dll file at install time I will take a look.
Does anyone have a benchmark comparison of code speed of Inline vs XS?
Peter Prymmer
