This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: -O1 -Winline


Mathieu Lacage <Mathieu.Lacage@sophia.inria.fr> writes:

| On Thu, 2004-09-16 at 02:06, James E Wilson wrote:
| > For C code, where inline 
| > tends to be used sparingly, and only for functions that are likely to be 
| > inlined, this warning may give you useful info.  For C++ code, where 
| > inline tends to be used a lot, and some things are implicitly inline, 
| > this warning will  just give you a lot of noise.
| 
| It would be great to add this to the gcc manual. Although what you say
| is obvious, I never thought about it before and, obviously, this poster
| did not either.

Well, people opinions vary on "inline" in C++ code.  Most certainly
when I say inline, I'm expecting the compiler to honor it in most
cases.  I don't buy the notion of implicit inline.  It is just plain
excuse and invention.  Inline functions in C++ started as being those
whose definitions appeared in-class definition.  It is only later that
the keyword appeared.  But, I guess that is just repeating a debate
that already took place. 

Some compilers have taken the history rewritting to the point where
they have completely transmuted the meaning of "inline", and propose
non-standard vendor lock-in "__inline" if one really means inline.
That is silly. It would be rather disappointing if GCC had to take
that route. 

-- Gaby


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]