Re: Updates of SE-PostgreSQL 8.4devel patches (r1403)

KaiGai Kohei wrote:
> Martijn van Oosterhout wrote:
> > On Tue, Jan 13, 2009 at 10:05:45AM -0300, Alvaro Herrera wrote:
> >> pgace.h: you have a bunch of "static inline" functions in here.  As far
> >> as I know this doesn't work in compilers other than GCC :-(  See
> >> pg_list.h (list_head) for an example.  I think we can tolerate this for
> >> the three functions in pg_list.h because they are so few and so tiny,
> >> but I'm not sure about PGACE because they are a large lot.  On the other
> >> hand, turning them to real functions would be a performance hit.
> > 
> > Really? C99 requires it and MSVC does support it. At least the other
> > compilers whose name I remembered (HP, Sun) support it also. I'd be
> > surprised if a compiler didn't since it's the form of inline that most
> > matches what people expect to happen.
> > 
> > Do you have an example?
> 
> I have no preference either of them, because it is not an essence of
> my patches whether its security hooks are implemented as inline, or not.
> 
> IIRC, indeed, some of compiler also supported "static inline".
> However, it also seems to me that PostgreSQL implementation tend to
> avoid to use inline functions actively.
> For example, heap_getattr() and fastgetattr() are implemented as
> macros, even if they have a bit complex conditional branches, which
> can be rewritten more simple with inline functions.

I thought one advantage of using macros is that we force the inlining,
while I think inline compiler directives are more of a hint, but maybe
the compiler knows better than we do in some cases.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +